Report

Child labour and

education
A survey of slum settlements
in Dhaka
Maria Quattri and Kevin Watkins

December 2016
Overseas Development Institute
203 Blackfriars Road
London SE1 8NJ

Tel. +44 (0) 20 7922 0300

Fax. +44 (0) 20 7922 0399
E-mail: info@odi.org.uk

www.odi.org
www.odi.org/facebook
www.odi.org/twitter

Readers are encouraged to reproduce material from ODI Reports for their own publications, as long as they are not being sold commercially. As copyright
holder, ODI requests due acknowledgement and a copy of the publication. For online use, we ask readers to link to the original resource on the ODI website.
The views presented in this paper are those of the author(s) and do not necessarily represent the views of ODI.

© Overseas Development Institute 2016. This work is licensed under a Creative Commons Attribution-NonCommercial Licence (CC BY-NC 4.0).

Cover photo: Abir Abdullah/ADB: Sohel works in an aluminium factory at Kamrangirchar in Dhaka.
Acknowledgements
This report was funded through a generous grant from the Bangladesh office of the United Kingdom’s Department for
International Development (DFID). Several DFID staff provided helpful comments on a draft proposal and methodology
note, including Golam Kibria, Mosharraf Hossain, Tayo Nwaubani and Afroza Chowdhury (Mimi).
Colleagues from BRAC Institute of Educational Development, BRAC University (BIED, BRACU) informed the design
of the survey and led implementation. The BIED, BRACU team comprised the following members: Md. Altaf Hossain,
Md. Abul Kalam and Sheikh Shahana Shimu. We are grateful for the insights, professionalism and dedication of BIED,
BRACU’s staff – and it was a privilege to work with them. In the course of the research many senior civil servants, NGO
representatives, parents, teachers, and business people were very generous with their time during interviews. They are too
numerous to mention. But we thank them collectively for the insights they provided.
We benefited from the advice and comments of many people. Special mention should be made of Furio Rosati,
Lorenzo Guarcello and Gabrielle Breglia at the Understanding Children’s Work (UCW) centre in Rome, the world’s
foremost institute on child labour analysis. Fazle Rabbani, Senior Education Specialist at the Global Partnership for
Education, provided advice on the overall approach and detailed comments on an earlier draft. Pauline Rose, Professor of
International Education at the University of Cambridge, also provided enormously helpful guidance and comments.
We are grateful to John Maher from Anaconda Editions Limited for editing the report under acute time pressure.
Mary Bose, now Executive Assistant to the Director at Save the Children, provided administrative support on project
management. Several ODI colleagues provided helpful advice, including Elizabeth Stuart, head of the Growth, Poverty
and Inequality programme and colleagues in the Public Affairs and Communications team.
Above all we want to thank the children living in Dhaka’s slums. We subjected 2,700 of them to in-depth interviews,
many of them after they had done a full day’s work. There is something profoundly humbling about engaging with
children on the wrong side of the opportunity divide. We spoke to 12-year-old girls working to put their siblings through
school, and to 10-year-old boys working so that their families had food on the table. Their resilience, optimism, and sense
of hope defy description, as does the magnitude of the national and international policy failures that destroy so much
human potential. We dedicate this report to them.

Child labour and education – a survey of slum settlements in Dhaka  1  

Contents

Acknowledgements 1

Summary 9

Introduction 11

1. Old wine, new bottles – child labour and education in the 21st century 13

1.1 The view from history 13

1.2 Measuring child labour 14

1.3 Wider literature review and evidence 15

1.4 Social protection and cash transfer programmes 16

1.5 Compulsory education and child labour prohibitions 17

2. Education and child labour in Bangladesh – strong progress and big challenges in an urbanising economy 19

2.1 Education for All – the unfinished agenda 20

2.2 Child work and child labour 25

2.3 Dhaka’s urban slums – the wider environment 28

3. Child labour and education in Dhaka’s slums 31

3.1 The Child Work and Education Survey (CWES) 31

3.2 Child labour – incidence, profile and characteristics 38

3.3 Some children are more at risk of child work than others 52

3.4 Schooling, learning and perceptions of education 55

Conclusions and policy recommendations 68

1 Building the evidence base 69

2 Making education a more effective part of the strategy for eliminating child labour 69

3 Linking education to strengthened child labour legislation and a more robust regulatory environment 70

References 71

Child labour and education – a survey of slum settlements in Dhaka  3  

List of boxes, figures and tables

Boxes
Box 1. The CWES in context 31

Box 2. Dhaka City Corporation – ‘North versus South’ 35

Box 3. Out of school and out of work 39

Box 4. Working lives  42

Box 5. The parents’ perspective  47

Box 6. Garment exports and child labour  48

Box 7. In-depth interview with the manager of a garment factory 49

Figures
Figure 1. Children’s employment and child labour – global estimates for 2012 15

Figure 2. Bangladesh is nearing universal primary school enrolment: reported enrolment rates between 2008–2015
by gender  21

Figure 3. Repetition rates are falling – but remain high: repetition by grade and gender, 2011–2014 21

Figure 4. Many children start school but do not complete a full primary cycle: drop-out rates by grade and gender, 2015  22

Figure 5. Attendance lags behind enrolment, with marked inequalities: enrolment and attendance rates,
selected groups in 2013 23

Figures 6a and 6b. Learning deficits remain large: learning achievement at Grades 3 and 5 for Bangla (Figure 6a)
and Maths (6b), 2013 24

Figure 7. Estimated child work and child labour in Bangladesh, 2013 25

Figure 8. National surveys point to an overlap between school and work: education status by age of children involved
in work, 2013 26

Figure 9. Thana survey locations 32

Figure 10. Share of population lacking access to safe sanitation in Bangladesh’s Thanas  33

Figure 11. Adult illiteracy in Bangladesh’s Thanas 34

4  ODI Report
Figure 12. Slum residency varies: average number of years the child’s household has lived in the slum 36

Figure 13. Slums have different patterns of migration: location from which the household moved prior to residency 36

Figure 14. Wealth and income disparities in Dhaka’s slums 37

Figure 15. School and work in Dhaka’s slums: household listing data on education and work status for children
aged 6–14  38

Figure 16. Age gradients for school and work status: children aged 6–14  39

Figure 17a. Working, schooling and out-of-school without work: reported status of boys (4,500 household listings)  40

Figure 17b. Working, schooling and out-of-school without work: reported status of girls (4,500 household listings)  40

Figure 18. Education and work status by slum location, children aged 6–14  41

Figure 19. Working hours: reported weekly work hours, children aged 6–14  42

Figure 20. Working lives start early: reported age of entry into employment for children who are working 43

Figure 21a. Reasons for working: top five reasons cited by boys  43

Figure 21b. Reasons for working: top five reasons cited by girls  44

Figure 22. Making choices: children reporting on who took the decision for them to start work?  44

Figure 23a. Cash payments dominate… 44

Figure 23b. …And most children have no contract  44

Figure 24a. Where are children working: source of employment reported by boys  45

Figure 24b. Where are children working: source of employment reported by girls  45

Figure 25. CWES slum locations and garment and apparel factories in Dhaka 46

Figure 26. The formal garment sector is reported as a major employer: age profile of 10–14 year-old children reporting
work in the formal garment sector 47

Figure 27. How is the child treated at work: perceptions reported by working children 49

Figure 28. Wage of child workers in the formal garment sector in comparison with national minimum wage
(in the garment industry) 51

Figure 29. The age gradient for working children: probability of working by age and gender for a child living in slums 52

Figure 30. Wealth matters: probability of working by age and wealth index for a child living in slums 52

Figure 31. More income is associated with less child labour: probability of working based on monthly household income
(excluding income from the child labour) 52

Child labour and education – a survey of slum settlements in Dhaka  5  

Figure 32. More settled residency patterns reduce child labour: probability of working based on the number of years
child household has been living in the same slum 53

Figure 33. Arrival from rural areas brings elevated risks of child work: probability of working based on
rural/urban relocation 53

Figure 34. Households with fewer children are less likely to have children working: probability of a child working
based on the percentage of children younger than 14 54

Figure 35. Father-headed households are more likely to include working children: probability of working based on
whether father or mother household head (HH) 54

Figure 36. Parental education strongly influences the child work patterns: probability of working based on the
education level of the child’s mother and father 54

Figure 37. Probability of working for a child living in slums of Dhaka City Corporation based on existing education
opportunities 55

Figure 38. Misalignment in age-for-grade profiles: share of children by age in Grades 1–8  56

Figure 39. Reasons for repetition: different groups of children attending and not attending school  58

Figure 40. Why do children drop out of school: children and parental perceptions by education and employment status  59

Figure 41. Repetition rates for in-school and out-of-school children by school type – percentage of children repeating
at least one grade 59

Figure 42. Reasons for school selection: parent (or guardian) responses on why they selected last school attended,
working children (last school attended) and non-working children  60

Figure 43. Spending on private tuition varies across households: reported spending by band, differentiated by
education/employment status of child  60

Figure 44. Accurate reading of a basic literacy sentence: percentage of children correctly reading the sentence
‘the girl is playing’ in Bengali  62

Figure 45. Testing for comprehension: share of children aged 11–14 correctly answering specified questions,
by education and employment status  63

Figure 46. School experience of children and interaction of school with parents: child’s response to learning difficulties,
understanding of textbooks and parent-teacher meetings, by employment/education status of the child 65

Figure 47. What is the ability level of children living in slums and can they succeed in education: responses of children
aged 6–14 by education and employment status 65

Figure 48. Future expectations: what children and parents say about their future employment, by employment and
education status 66

6  ODI Report
Figure 49a. How often do you feel sad/unhappy? 67

Figure 49b. My future is… 67

Figure 49c. How do you feel about yourself? 67

Tables
Table 1. Primary gross and net attendance rates: slums versus the urban average, 2011 23

Table 2. Dhaka City Corporation slum areas and population 29

Table 3. Sample selection: 23 parts and sub-parts in 38 administrative units 34

Table 4. Education and employment status by age cohort: children aged 6–10 and 11–14 38

Table 5. Children reporting compulsion in the workplace, selected practices 49

Table 6. Percentage of children that experienced any of the following… 50

Table 7. Reported income by household status: average income levels for mother, father and working children 50

Table 8. Reported income by household status: average income levels reported by child (Bangladeshi takas, (US $)) 51

Table 9. The education profile of working children aged 11–14 years 56

Table 10. Highest reported grade for working children aged 10-14 57

Table 11. The education profile of non-working children who are out-of-school aged 11–14 57

Table 12. Patterns of grade repetition 57

Table 13. Schools attended by type and child status: in school, out-of-school and work, and child labourers,
ages 11–14 58

Table 14. Reading for basic literacy: correct identification of 10 letters in Bengali alphabet by age and work/education
status: share of children in age groups 6–10 and 11–14 61

Table 15. Reading for basic literacy: correct identification of five words in Bengali by age and work/education status:
share of children in age groups 6–10 and 11–14 61

Table 16. Testing for reading: outcomes for children aged 6–10 and 11–14, proportion able to read and time taken 63

Table 17. Numeracy and maths competencies: correct results by age group and education/employment status,
children aged 6–10 and 11–14 64

Table 18. Reported classroom experiences of children aged 6–14 by employment and education status:
reporting on perceptions of teachers 64

Child labour and education – a survey of slum settlements in Dhaka  7  

Table of Abbreviations

BANBEIS Bangladesh Bureau of Educational Information and Statistics

BBS Bangladesh Bureau of Statistics
CDL Child domestic labour
CLS Child Labour Survey
CWES Child Work and Education Survey
CREATE Consortium for Educational Access, Transitions and Equity
DCC Dhaka City Corporation
DHS Demographic and Health Surveys
FRAs Field research assistants
IPEC International Programme on the Elimination of Child Labour
ILO International Labour Organization
MICS Multiple Indicator Cluster Survey
NGO Non-governmental organisation
ODI Overseas Development Institute
PCA Principal Components Analysis
PPS Probability Proportionate to Sample Size
PSU Primary Sampling Unit
UCW Understanding Children’s Work
UNICEF United Nations Children’s Fund

8  ODI Report
Summary
Universal secondary education and quality learning have
been adopted by the international community as goals for
2030. Child labour, which remains endemic in many poor
countries, represents an obstacle to both goals. It keeps
children out of school, hinders effective learning and denies
children an opportunity to acquire the knowledge and
skills they need to escape poverty, and that their countries
need to drive inclusive growth and human development.
This report describes a unique, large-scale representative
survey into the relationship between child labour and
education in slum settlements located across eight
Thanas (administrative units) in Dhaka, the capital of
Bangladesh – a city and a country in which child labour
remains widespread. The survey covers over 2,700
households drawn from a listing of 4,500, with children
themselves as well as their parents responding to a detailed
questionnaire. It is representative for a population of just
over half-a-million people.
The backdrop to the survey is Bangladesh’s strong
record on human development. The country has registered
extraordinary advances in education, notably on access
and gender equity. Poverty levels have fallen with strong
and relatively inclusive economic growth. Looking ahead,
Bangladesh faces immense challenges. While the country is
nearing universal primary school enrolment, drop-out rates
remain high – and learning levels are poor. Urbanisation
has been one of the engines of Bangladesh’s development.
But it has gone hand-in-hand with the rapid growth of
urban slums marked by high levels of poverty and low
levels of service provision. Child labour is rife in these
slums. While our survey is not nationally representative,
it provides a window on the world of child labour in the
megacity of Dhaka, which accounts for almost half of
Bangladesh’s slum population.
We report a number of findings that differ from those
to emerge from previous research and national surveys.
Five themes stand out. First, there is a high work incidence
among children aged 6–14, with an average rate of 15%
reported across our survey sites. An equivalent proportion
is neither in school nor working. Second, in contrast to
previous studies, we find a high level of segmentation
between work and school: the overwhelming majority of
working children are out-of-school. Third, the distinction
between child work and child labour is of questionable
relevance in Dhaka’s slums. Almost all working children
are involved in what both International Labour
Organization (ILO) Conventions and national legislation
would categorise as child labour, with hazardous work
figuring prominently. Fourth, child labour markets in
Dhaka are highly segmented by age and gender. Boys make
the transition to the world of work earlier and in greater
numbers than girls, with labour market entry taking off
from the ages of 10 and 11 respectively. Child labour levels
rise from around 8% at the age of 10 to 45% at the age of
14. Fifth, the ready-made garments sector appears to be a
major employer of children, accounting for two thirds of
female child labour.
Wider findings from the research illustrate the damaging
interaction between child labour and education in
Bangladesh. Children are driven into work by economic
hardship – and we find evidence that wages from child
labour equalise average income across slum households
with and without working children. This lends weight
to the view that parents make the decision to send their
children into work as a ‘distress choice’ aimed at meeting a
minimum income level.
The education system contributes to the child labour
problem by creating a ‘supply’ of potential recruits.
Child workers enter labour markets having accumulated
limited years of schooling. On average they make the
transition to the world of work with just four years of
schooling – and the average 14-year-old child labourer
has not completed Grade 3. Basic literacy and numeracy
skills are poor for all children, including those in school.
But they are worst for child labourers and children who
are neither in school nor working. Late entry to school,
grade repetition and poor quality education all serve to
push children out of education and into employment. More
effective enforcement of compulsory education legislation,
coupled with improvements in the quality of schooling and
measures to counteract the effects of household poverty,
could accelerate progress towards the eradication of child
labour in Dhaka’s slums. However, while current policies
set ambitious goals, their effectiveness is hampered by
the absence of an integrated strategy for education, the
reduction of child poverty and enforcement of child labour
laws.
One of the take-home messages to emerge is that
Bangladesh will not achieve the 2030 development goals
on education and other objectives without a strengthened
commitment to eradicate child labour; and the country
will not eradicate child labour without making education
compulsory and free for the 6–14 age group.
As in other countries, child labour in Bangladesh is
a complex phenomenon that has to be addressed in a
coordinated fashion on a cross-sectoral basis. Effective
Child labour and education – a survey of slum settlements in Dhaka  9  
action requires integrated approaches that span the
regulation of labour markets, education, child welfare and
wider strategies for poverty reduction. It also entails a
process of dialogue aimed at changing attitudes. Based on
the findings from our survey, we set out a number of key
recommendations. These include:
•• Developing an improved evidence base to capture the
extent of child labour and the education status of child
labourers, with a focus on urban slums and surveys that
capture the reported experiences of children.
•• Making the eradication of child labour a central
objective of education policy.
•• Raising the age for free and compulsory education from
10 to 14.
•• Increasing overall financing for education to 4–5% of
GDP, with a greater emphasis on provision in slum areas
and overall equity.
•• Improving the quality of education in slum areas
and enhancing the school readiness of slum-dwelling
children through universal early childhood provision in
slums.
10  ODI Report
•• Supplementing current education stipend programmes
through a programme to provide cash transfers at
primary and junior levels to slum-dwelling children –
the timing and level of the transfer should be calibrated
against current monthly earnings ($47 a month on
average).
•• Strengthening the regulatory environment for child
labour by ratifying ILO Convention 138, reducing the
current threshold for ‘hazardous work’ from its current
level of 42 hours, and expanding the list of prohibited
hazardous occupations to be consistent with ILO
standards.
•• Investing in the human resources needed to ensure that
regulatory agencies are equipped to conduct effective
inspections.
•• Imposing more punitive fines on employers found to be
employing under-age workers.
•• Reviewing inspection arrangements for the garment
sector to ensure that factories comply with national
laws.
Introduction
Child labour remains endemic in many of the world’s
poorer countries. According to the ILO, 168 million
children aged 5–17 are working as child labourers –
around 12% of all children in that age group. These
children are not just working for a few hours a day.
They are engaged in activities that compromise their
opportunities for education. Half of them are in
‘hazardous’ occupations that pose risks for their health,
safety and security (ILO et al., 2013). While the ILO
reports that child labour numbers have declined since
2000, the numbers are still distressing. Millions of the
world’s children continue to live and work in conditions
that would have been familiar to children in Europe and
the US some two centuries ago.
The persistence and scale of child labour represents a
barrier to the achievement of the Sustainable Development
Goals (SDGs) set for 2030. These goals include the
eradication of poverty, decent quality learning for all
children up to secondary school level, reduced inequality
and the creation of decent jobs.1 Mass child labour is
inconsistent with these targets, notably those associated
with education (Guarcello et al., 2015). For the children
directly affected, early entry into the world of work
represents a loss of freedom, a violation of rights, a source
of vulnerability and a constraint on learning. Apart from
exposure to risk of injury, these children are denied a
chance to acquire what Amartya Sen describes as ‘human
capabilities’ – the knowledge, skills and competencies
needed to expand choice and extend opportunity.
Children making early entry into the world of insecure,
unskilled, low-paid work are unlikely to accumulate the
education they need to secure decent work and break
the transmission of poverty across generations. For
countries, too, extensive child labour is a roadblock to
human development because it erodes the human capital
on which dynamic and inclusive economic growth,
rising productivity and social progress depend. It is no
coincidence that child labour is strongly associated with
low income both across countries and within them.
There is a strong link between educational disadvantage
and child labour. Children forced out of school and into
work to help their families make ends meet face elevated
1 The SDG targets for education envisage a world in which ‘all girls and boys complete free, equitable and quality primary and secondary education leading
to relevant and effective learning outcomes’, with an overarching injunction ‘to reach the furthest behind first’ (United Nations General Assembly, 2015).
Widespread and persistent child labour represents an obvious obstacle to these ambitions.
2 BRAC is the world’s largest development organisation, dedicated to empowering people living in poverty.
risks of getting trapped in a cycle of deprivation. The
transmission operates in both directions. The interaction
between poverty and labour markets pulls children into
child labour. But ‘failure’ in education can generate a push-
factor by creating a supply of potential child labourers
from the ranks of children whose school careers have been
prematurely ended.
There is a consensus that early exposure to work
and withdrawal from education are harmful – and that
international cooperation and national legislation should
seek to restrict child labour. However, the limited progress
made towards a world free of child labour is indicative
of the resilience, complexity and depth of the underlying
social disadvantages and power relationships that
perpetuate it.
This report makes a contribution to the wider debate
on child labour and education. It focuses on Bangladesh.
This is a country which, over the past three decades, has
made extraordinary progress on a wide range of human
development indicators, notably education. Governments
have put in place wide-ranging strategies for eradicating
child labour. Yet the practice continues at endemic levels.
Failure to tackle this will jeopardise Bangladesh’s prospects
for accelerating progress towards the 2030 development
goals.
We present the results of a survey conducted in slum
settlements in Dhaka, Bangladesh’s capital city. The survey
was developed and implemented by BRAC2 Institute
of Educational Development, BRAC University (BIED,
BRACU) and the Overseas Development Institute (ODI).
The survey is representative for a population of over half-
a-million people across eight Thanas in the Dhaka City
Corporation (DCC) area. The survey sample was derived
from a listing of 4,500 households, from which 2,700
slum-dwelling children and their parents were randomly
identified and interviewed. To our knowledge, it is the
largest survey of its type with a distinctive focus on child
labour and education to have been conducted in Dhaka, or
in Bangladesh as a whole.
We caution against drawing national conclusions from
a survey that is representative for urban slum locations in
one city. However, these locations are home to the largest
Child labour and education – a survey of slum settlements in Dhaka  11  
slum populations of children out of school in Dhaka, one
of the world’s fastest-growing megacities in a country that
is rapidly urbanising. The picture that emerges from the
survey may have a wider relevance to an understanding
of the complex relationship between urbanisation,
child labour and human development. Looking beyond
Bangladesh, our survey raises wider issues of relevance
to the 2030 development goals. If the international
community is intent on ‘leaving no one behind’ and
reaching ‘the furthest behind first’, as envisaged in
the goals, breaking the link between child labour and
educational disadvantage would appear an obvious
starting point (United Nations, 2016).
The rest of the report is structured as follows. Part
1 provides an overview of the concepts that underpin
distinctions between child work and child labour, along
12  ODI Report
with a brief review of the wider theoretical literature on
child labour and education. Part 2 sets the national context
for our survey. It examines what we describe as a ‘twin
crisis’ in education spanning access and learning, and
summarises some of the wider evidence on child labour
to emerge from national studies. Rapid urbanisation is
one of the defining feature of Bangladesh’s development
and Dhaka has emerged as one of the world’s megacities.
We explore the place of slums in the country’s urban
development. Part 3 reports on our survey findings. We
provide a picture of the incidence of child labour (drawing
on our household listing sample), profiles of working
children and an account of the interaction between child
labour and education. Part 4 sets out what we see as some
of the critical lessons to emerge and recommendations on
policy.
1. Old wine, new bottles –
child labour and education
in the 21st century
The elimination of child labour is a contemporary concern
with a long history. During the first half of the 19th
century, child labour was a common feature of the social
landscape in today’s advanced economies. The debates
among political leaders, social reformers and industrialists
have a powerful contemporary resonance. One of the
recurrent themes was the drive to set progressively lower
limits on the hours that children spent in work, while
gradually increasing the time spent in education. In this
section we look at how moves towards the eradication of
child labour have been shaped by shifting perceptions of
childhood – and at how approaches to the measurement of
child labour have changed over time.
1.1 The view from history
Countries such as Bangladesh are still in a state of
transition away from mass child labour and towards
compulsory schooling. Many of the choices facing
parents and the experiences of children would have been
familiar to parents and children in 19th-century Britain.
Here are two brief excerpts from a British parliamentary
investigation – the Sadler Committee – into child labour in
the textile industry in 1832, the first is with Joshua Drake,
the parent of a working child:
Q Why do you allow your children to go to work at
those places where they are ill-treated or over-worked?
A –Necessity compels a man that has children to let
them work.
Q Then you would not allow your children to go to
those factories under the present system, if it was not
from necessity?
A –No.
Q Supposing there was a law passed to limit the hours of
labour to eight hours a day, or something of that sort,
of course you are aware that a manufacturer could not
afford to pay them the same wages?
A –No, I do not suppose that they would, but at the
same time I would rather have it, and I believe that it
would bring me into employ; and if I lost 5d. [English
fivepence] a day from my children’s work, and I got
half-a-crown myself, it would be better.
The second is with a former child labourer named
Matthew Crabtree:
Q At what age did you first go to work in a factory?
A –Eight.
Q Will you state the hours of labour at the period when
you first went to the factory, in ordinary times?
A –From 6 in the morning to 8 at night.
Q With what intervals for refreshment and rest?
A –An hour at noon.
Q During those long hours of labour could you be
punctual; how did you awake?
A –I seldom did awake spontaneously; I was most
generally awoke or lifted out of bed, sometimes
asleep, by my parents.
Q What was the consequence if you had been too late?
A –I was most commonly beaten.
Q When you got home at night after this labour, did you
feel much fatigued?
A –Very much so.
Q Had you any time to be with your parents, and to
receive instruction from them?
A –No.
Like many other parents, Joshua Drake allowed his child
to work as a consequence of economic hardship. Historians
have documented the grave concerns of mid-19th century
English parents over the decision to allow their children
to work (Horrell and Humphries, 1995). What is striking
about the first exchange is that it addresses a question that
has become the staple of labour market economic analysis
in child labour debates: namely, would a regulatory limit
on child labour create employment opportunities for the
parent (Baland and Robinson, 2000; Basu and Van, 1998).
The second respondent powerfully captures the invasive
and destructive impact of extreme child work on the
well-being of children – and the child’s awareness of the
consequences.
Child labour and education – a survey of slum settlements in Dhaka  13  
 There is another area in which the proceedings of the
1832 enquiry have a contemporary resonance. The Sadler
Committee played an important role in shaping legislation,
introduced the following year, to limit to eight hours the
time children could spend working in garment factories.
Some 190 years later, our survey of slum dwellers in
Dhaka, Bangladesh, found children reporting an average
working time of 10 hours a day.
1.2 Measuring child labour
Debates on child labour are shaped by national contexts,
history, politics and the shifting sands of values and norms.
How the terms ‘child’, ‘work’ and ‘labour’ are defined
has a critical bearing on any estimate of the extent of the
underlying practice – and on policies to curtail or prohibit
child labour.
1.2.1 Child labour and child work – defining the
boundaries
Child labour is a legal concept with powerful ethical
undertones rather than an objective condition easily
captured by statistics. Translating legal norms into cross-
country data is inherently problematic. The standards set
by ILO Conventions represent an international reference
point – and the source of the headline figures cited in
the Introduction (Diallo et al., 2013; Edmonds, 2008) .
However, the headline statistics provide a static picture of
what is a dynamic and highly differentiated condition.
For national and international reporting purposes, child
labour is different to child work. Child labour represents
a subset of employment or working activities. Three
Conventions define the legal boundaries that separate child
labour for elimination from ‘legitimate’ child work. ILO
Convention 138 (1973), provides the most comprehensive
and widely adopted international definition of the
minimum age for admission to work (ILO, 1973). Ratified
by 169 countries, it calls on states to set a threshold of at
least 15 years of age as the minimum for entry to work,
with a provision for 14 years of age in less developed
countries. For hazardous work it raises the age threshold to
18. Hazardous work covers a range of sectors (mining and
construction, for example) and occupations that run from
a range of metallic and electrical engineering activities to
operating textile, wood-cutting, leather product machinery,
and street vending.
3 Other non-signatories include India and a range of OECD countries, among them the US, Australia and New Zealand.
4 India is one of the countries that has not ratified the Convention. The ILO designates hazardous occupations and sectors (see ILO, 2012).
5 Economic activity covers all market production and certain types of non-market production, principally the production of goods and services for own use
(Edmonds, 2008).
14  ODI Report
Convention 138 sets out parts of the boundary that
separates child work from child labour. It allows children
from the age of 13 (12 in less developed countries) to
engage in light work which is not harmful to their health
and does not ‘prejudice their attendance at school … or
training programmes’. Permissible light work is defined as
any non-hazardous work by children that amounts to a
total of less than 14 hours a week. Only 18 countries have
not ratified the Convention.3 Bangladesh is one of them.
The second boundary-setting international instrument
is ILO Convention 182 (ILO, 1999) on the worst forms
of child labour. This calls on states to prohibit through
‘immediate action’ all forms of slavery and compulsory
labour, the involvement of children in illicit activities and
work which ‘is likely to harm the health, safety or morals
of children’ (ILO, 1973). The Convention has been ratified
by 180 countries, including Bangladesh.4
The Convention on the Rights of the Child is the
third legal pillar – and the most comprehensive child
rights treaty. It explicitly recognises the right of the child
‘to be protected from economic exploitation and from
performing any work that is likely to be hazardous or to
interfere with the child’s education, or to be harmful to
the child’s health’. The Convention also calls on states to
provide compulsory and free education (Article 28). An
important aspect of the Convention is that it establishes
children as ‘rights holders’ and requires parents and other
adults to act in the best interests of the child (Article 3),
and governments to enforce child rights through national
action and international cooperation (Article 4).
Implicit in each of these Conventions is a recognition of
the strong link between child work and education. Indeed,
the boundaries for legitimate work are defined in part by
the requirement that children should be able to develop
their potential through education.
All of this serves to underscore the hazards of estimating
child labour. The ILO distinguishes between children’s
employment in ‘economic activity’ for at least one hour
a week and child labour, as defined under the core
Conventions.5 In simplified form, children’s employment
can be thought of as the envelope that encompasses
legitimate forms of child work, and both non-hazardous
and hazardous child labour (Figure 1).
 Translating broad international legal norms into
legislation and statistical measurement tools is intrinsically
difficult. Despite ILO Convention 138, different countries
have variable standards for the minimum age of work.
They also have very different definitions of ‘light work’
and different interpretations of when work interferes with
schooling. In the case of Bangladesh, children aged 12
years and above are permitted to engage in ‘light work’
for up to 42 hours a week – a level well in excess of the
Convention 138 standard. In addition, there are divergent
definitions of hazardous work, with national lists often
diverging from the ILO’s hazardous work list.
Figure 1. Children’s employment and child labour – global
estimates for 2012
Children in
employment
264,427,000
Child labour
167,956,000
Hazardous child
labour
85,344,000
Data source: ILO et al. (2013) Marking Progress Against Child
Labour: Global estimates and trends 2000-2012. Geneva:
International Labour Office.
In the absence of comparable cross-country survey data
of high quality and in the face of limited country coverage,
global estimates are derived through an elaborate process
of extrapolation and interpolation. Misalignment between
international definitions of child labour used by the ILO
and national legal definitions can produce wide variations
(Khan and Lyon, 2015). Indeed, estimates produced by
national authorities can vary widely dependent on the
definitions of child work and child labour employed.
This backdrop has an important bearing on our survey
in Bangladesh. Identifying which aspects of child work
constitute ‘child labour’ for elimination looks different
from the perspective of national legislation in the country,
than from the perspective of ILO Conventions. However,
from the perspective of the children living and working
in the (admittedly very distinctive) slum environments
we survey, some of the standard legal distinctions drawn
between child work, child labour and hazardous labour are
of questionable relevance. What is clear from our survey
results is that the spirit and the letter of the injunctions
set out in ILO Conventions, the Convention on the Rights
of the Child and national legislation on child labour and
education produce limited results in the real world of many
children in Dhaka’s slums.
1.3 Wider literature review and evidence
Child labour is overwhelmingly concentrated in poor
countries and, within those countries, among poor
households. This provides a useful insight into the
mechanisms linking poverty and social deprivation to child
labour. Parents of working children in early 19th-century
Britain were unhappy about the choices forced on them,
as the proceeding of the 1832 parliamentary investigation
cited earlier make clear (Horrell and Humphries, 1995;
Cunningham and Viazzo, 1996). The parents of 21st-
century child labourers living in the slums of cities like
Dhaka or Jakarta, or poor rural areas in Ghana and India,
are no different. They are similarly compelled by necessity
to sacrifice their children’s education and well-being not
out of free choice, but economic compulsion.
1.3.1 Theoretical models on household
decision-making
There is an expansive literature on the relationship
between child labour and education. Much of that
literature derives from theoretical models for household
decision-making and labour market participation. More
recently, this literature has been supplemented to empirical
work on transitions to work and the relationship between
employment and schooling. Both sets of literature are
relevant to the analysis of child labour in Bangladesh.
Much of the theoretical analysis of child labour can
be traced back to Gary Becker’s work on intra-household
bargaining (Becker, 1964; Rosenzweig and Evenson, 1977).
In Becker’s unitary household model, child labour is the
product of unequal bargaining relationships. With children
enjoying limited bargaining power, parents and employers
effectively bargain over children’s wages and the share
of wages to be paid as food. What shapes decisions over
whether to send children to school is the relative power of
the household in relation to the employer.
Another strand of literature has addressed child labour
from a human capital approach (Baland and Robinson,
2000). Putting children to work in this perspective
increases the basket of goods a family can consume in the
short run. But there is a potential trade-off between current
and future income, with sub-optimal social and economic
returns over the long run. When children have to work,
they risk losing out on education. Many child labourers
either never attend school or they drop out early, while
those struggling to combine school and work often register
lower levels of learning.
Child labour and education – a survey of slum settlements in Dhaka  15  
 The interplay between child labour and education has
been extensively explored (Guarcello et al., 2015; UCW,
2015; de Hoop et al., 2016). The dynamic interactions
between household poverty and parental preferences
on the one side and school-related factors – such as the
perceived or actual quality of schooling and accessibility
of schools – is complex. What emerges clearly from the
research, however, is the tension between child labour and
education targets of the type adopted under the SDGs for
2030.
If children working is the problem, is a prohibition on
child labour the solution? That question has been at the
centre of a substantial literature since the seminal papers of
Basu and Van (1998), and Basu (1999). Starting from the
proposition that parents do not voluntarily or maliciously
send their children to work, but do so in order to achieve
a subsistence income, an obvious issue arises: would the
prohibition of child labour produce more of the poverty
that drives children into child labour?
Theoretical modelling produces two main predictions.
The first is that the principal determinant of parental
decisions to send children to work are poverty and
imperfect credit markets, which prevent poor households
from borrowing to realise future social and economic
returns (Basu and Van, 1998; Grootaert and Kanbur, 1995;
Baland and Robinson, 2000).
The second prediction is that the effect of a ban on
child labour will depend on underlying market conditions.
If children are withdrawn from labour markets and
their wages rise to maintain household incomes at the
same level, the ban could move the economy to an
equilibrium without child labour (Basu, 1999). However,
in an economy where many children are working, poor
households are heavily reliant on child labour, and adults
are unable – or unwilling – to substitute for children
(perhaps because they are already fully employed), a
prohibition can hurt the poorest households by keeping
wages low. Outcomes depend, in part, at least on the
response of firms to changing labour market equilibria
(Soares, 2003).
Recourse to child labour in the underlying model is
seen as an act of desperation. Parents might be aware of
the social and economic costs of sending children to work,
but adult wages and employment prospects are insufficient
to meet a defined subsistence level with child labour. Our
survey in Dhaka lends some weight to this interpretation
(see Section 3).
1.3.2 Empirical evidence backs the theory
Empirical research on the effect of external consumption
shocks confirms the transmission mechanisms operating
between poverty and child labour. Panel data analysis
for Tanzania found that credit-constrained households
responded to transitory income shocks caused by crop
losses, by increasing child labour and reducing education
in order to smooth consumption. Research on the response
16  ODI Report
to the 1998 floods in Bangladesh found that child labour
increased not just in relation to the scale of the shock,
but also in relation to the availability of credit (Alvi and
Dendir, 2011). These are illustrations of market failures
in areas such as risk insurance, capital and borrowing
(Grootaert and Kanbur, 1995).
Children from poorer households are consistently more
exposed to the risk of child labour – and to the educational
disadvantages that come with child labour. Evidence to
this effect comes from all regions and a growing number
of countries (Guarcello et al., 2015; UCW, 2015). In the
Philippines, 9% of children from the poorest background
are involved in child labour compared to 3% for the
richest (Guarcello et al., 2015). Unsurprisingly, working
children often struggle to combine employment and
education. One review of evidence from 19 countries found
that school attendance rates for children in employment
were consistently lower than for non-working children
(Guarcello et al., 2015). While a large share of working
children do attend school, there is evidence that they
lag behind their non-working peers in terms of grade
progression (Khan and Lyon, 2015; Guarcello et al., 2015;
UCW, 2015). The effects increase in magnitude with hours
of work.
1.4 Social protection and cash transfer
programmes
If poor households send children to work because of cash
constraints and credit market imperfections, increased
income and risk insurance might be expected to weaken
the link between poverty, child labour and educational
deprivation. An increasing body of evidence from cash
transfer programmes points in precisely this direction.
Parental decision-making on school attendance is the
flip-side of decision-making on labour market entry. Simple
economic models for understanding school participation
and household investment in education assume that
parents seek to maximise life-cycle utility for their children.
The decision on whether or not to send children to school
will be based on perceived costs and benefits. However,
parents may underinvest relative to socially optimal levels
for a number of reasons, including imperfect information
on the benefits of education, poverty-related credit
constraints, and differences between individual and socially
optimal returns to education.
Cash transfer programmes can affect child labour
by changing the propensity to attend school. These
programmes can increase returns to keeping children
in school, reduce returns to child labour and enable
households to smooth consumption in the face of
exogenous shocks. By providing resources to the household
they relieve poverty, lower risk and mitigate market
imperfections limiting credit, thereby making it possible
for households to afford more education (and forego more
child labour). Programmes which condition payments on
school attendance create an incentive effect by increasing
the immediate returns on children being in school and
decreasing returns on child labour. Precise transmission
effects will be determined by design factors, including
the level and timing of transfers, and whether or not the
transfers are made conditional on children attending
school. De Hoop and Rosati (2014) provide a theoretical
framework for understanding how cash transfers may
modify household decision-making.
Conditional cash transfer (CCT) programmes in
education have been extensively evaluated (Glewwe
and Muralidharan, 2015; Bastagli et al., 2016). Among
the consistent themes to emerge is a positive effect
on school attendance, with supplementary benefits
for grade progression and completion. One literature
review covering 43 impact estimates associated with
cash transfer programmes from 20 high-quality studies
found that almost all (40 in total) were positive (Glewwe
and Muralidharan, 2015). The studies cover a range of
contexts. Large national programmes such as Progresa
in Mexico (Behrman et al., 2009, Behrman et al., 2011)
and Bolsa Família in Brazil (Glewwe and Kassouf, 2012)
have registered marked gains in school attendance, grade
progression and transition from primary to secondary
school. In Malawi, a CCT for girls increased daily
attendance rates by eight percentage points over two
years (Baird et al., 2011). An unconditional cash transfer
programme – the Social Cash Transfer Scheme – similarly
led to a small, but significant, rise in enrolment over the
next two years for girls already in school (Innovations
for Poverty Action, 2011). Evaluations of Nicaragua’s
Red de Protección Social found significant increases in
enrolment among children aged 7–13 (Gitter and Barham,
2008). Another evaluation using a randomised selection
of communities – both covered and not covered by the
programme – found an increase in the number of grades
completed (Barham et al., 2013). This is consistent with
wider research that has found a reduction in the aggregate
level of child labour among poor rural households
receiving cash transfers (Del Carpio and Loayza, 2012).
Introduced on a randomised control trial basis, Honduras
provided grants to households conditional on children
reporting 85% attendance: enrolment increased by eight
percentage points (Galiani and McEwan, 2013).
None of this amounts to automatic evidence that
improved school attendance is the result of reduced child
labour. Households could consume or invest the cash
transfer they receive (potentially drawing children into
work) while seeking to maintain child work at previous
levels, perhaps by reducing the leisure time available to
children. This would call into question the assumptions
about household decision-making outlined earlier.
Results of high-quality peer reviewed evaluations do
not point in this direction. One review of 19 studies of
cash transfer evaluations that include estimates for transfer
impacts on child labour found eight that registered any
significant impact. All showed a decrease in child labour.
Another five studies analysed in the review report on
impacts related to the intensity of child labour, as measured
by hours worked. All five found statistically significant
reductions, ranging from 0.3 hours to 2.5 hours a week
(Bastagli et al., 2016). While there are marked variations
in impact ranges reflecting the size, scale and mode of
delivery for the transfer, the evidence supports a conclusion
drawn by de Hoop and Rosati (2014): ‘the effects of any
household investments in productive assets and activities
that draw children into work are offset by stronger income
and substitution effects that keep children in school and
out of work’.
1.5 Compulsory education and child labour
prohibitions
In today’s advanced economies, the demise of mass child
labour was associated with the introduction of legislation
making education compulsory. Did a combination of
prohibition on children working and compulsory schooling
play the central role in consigning child labour to history?
And are there any lessons from late 19th-century and early
20th-century Europe and the US that might have relevance
for policy-makers in countries like Bangladesh?
These questions have figured prominently in debates
on child labour (Cunningham and Viazzo, 1996; Basu,
1999). To the extent that any lesson can be drawn, it is
that the evidence cautions against looking for cure-all
solutions. In Britain, the introduction and enforcement of
compulsory education in 1880 did contribute to a decline
in child labour, along with earlier factory acts legislating
for children to spend more time in schooling (Cunningham
and Viazzo, 1996). Raising the school leaving age from
10 in 1880 to 14 in 1918 also made a difference, notably
by redefining childhood. Once again, comparisons with
Bangladesh are instructive. Compulsory education in the
country today has the same duration as in the Britain of
1880.
Evidence from the US points in a different direction. The
expansion of state compulsory schooling and anti-child
labour laws in the four decades from 1910 was associated
with a sharp increase in secondary school participation,
from 18% to 71% (Goldin and Katz, 2003). However,
association is not causation. Legislation is estimated to
have accounted for no more than 5% of the increase in
enrolments. Increased incomes, changed attitudes and a
growing recognition of the importance of education were
the primary drivers of change (Goldin and Katz, 2003).
Prohibition, compulsory schooling and rising living
standards interacted to create a virtuous circle of more
schooling and less child labour across the industrialised
world. The precise mix of forces varied across countries,
time periods and contexts. Legislative restrictions on child
labour and the eventual declaration of child labour as
illegal, were both a cause and effect of changing social
Child labour and education – a survey of slum settlements in Dhaka  17  
attitudes (Hopkins, 1994; Brewer and Porter, 1994;
Humphries, 2012). At the same time, it was increasing
prosperity that made it possible to pull children out of
work without households being consigned to poverty
(Basu, 1999; Cunningham and Viazzo, 1996).
For a country like Bangladesh there are two broad
conclusions to be drawn from our survey of slum-dwelling
children. The first relates to legislation. Extending
compulsory education to the age of 14 is likely to prove
more effective than enforcing a ban on child labour
through factory and informal sector inspection, though the
latter certainly needs strengthening (see Section 2). Apart
from anything else, monitoring a child’s presence in school
is administratively easier than monitoring (and potentially
prosecuting) small-scale employers.
The second conclusion is that neither compulsory
education nor prohibition is a stand-alone solution.
The evidence on the impact of child labour bans in
developing countries, whether linked to compulsory
education or not, is limited. Researchers have looked
at this issue through analysis of the effects of minimum
employment age legislation. Evidence from Brazil (relating
to 1998 legislation which increased the minimum age
of employment from 14 to 16) points to a modest
decline in child work, mostly on the part of boys, and
a corresponding increase in in the proportion of boys
18  ODI Report
studying (Piza and Portela Souza, 2016). In Pakistan, the
1991 Employment of Children’s Act similarly appears to
have lowered child employment, though not in agriculture
(Fasih, 2007). Another study found that similar legislation
in India had the perverse effect of increasing child
labour, reducing wages and lowering school attendance
(Bharadwaj et al., 2013). Cross-country research has
found no consistent effects. One study on child labour and
schooling drawing on micro-data from 60 predominantly
low-income countries found no short-term effects from
legislation on minimum working age.
While the evidence base remains limited, it is relevant
for policy-makers in Bangladesh. Policy outcomes are
clearly conditioned by the wider economic context and
labour market conditions – and by the wider policy
environment. Introduced as a stand-alone measure in a
situation where less child labour will exacerbate poverty
and increase vulnerability, the likely result is that parents
will allow their children to find illicit routes back into
work. If the alternative to work is the prospect of hunger,
legislation on compulsory schooling will produce limited
results. However, coupling that legislation with cash
transfer programmes, targeted support, the provision of
good quality schools, and interventions that improve the
labour market position of the poorest households, has the
potential to produce transformative results.
2. Education and child
labour in Bangladesh
– strong progress and
big challenges in an
urbanising economy
This section provides a brief overview of the intertwined
challenges facing Bangladesh in education and child labour.
It provides a bird’s eye view of the progress achieved
and the challenges ahead. These challenges have to be
addressed against a backdrop of rapid shifts in human
geography, as Bangladesh becomes an increasingly urban
society.
Bangladesh’s achievements have been rightly lauded.
The country is among the top ranking development success
stories of the Millennium Development Goal era. Many
of the targets set for 2015 – on child mortality, maternal
mortality, education, gender equity and poverty – were met
well ahead of schedule. Strong and broad-based economic
growth has driven a respectable rate of poverty reduction.
Growth has averaged over 6% since 2000, with per capita
incomes doubling to $3,191 in 2011 Purchasing Power
Parity (PPP).6 The national poverty rate fell from 49%
in 2000 to 31% in 2010, while extreme poverty rates
halved over the same period.7 Measured on the Human
Development Index scale, Bangladesh ranks 142 out of 188
countries, but it outperforms countries – including India
– at higher levels of per capita income on key measures of
progress (UNDP, 2015).
The progress report does not diminish the scale of
the challenges ahead. Bangladesh remains a low-income
country. Millions of the country’s citizens live just above
the poverty line: in 2010 there were 50 million living on
between $1.25 and $2 a day. While income inequality
6 Data are from Centre for Policy Dialogue (URL: http://cpd.org.bd/).
7 The extent of poverty was also halved, from 34% to 17%. Measured on the old international poverty line of $1.25 (at 2005 PPP), 43% of the population
was estimated to be living below the threshold in 2010, compared to 58% in 2000 (World Bank, 2013).
remains modest by global and regional standards – the
Gini coefficient in 2010 was 30 – social disparities are
increasingly visible, not least in cities were slums are
located in close proximity to increasingly affluent areas.
Employment is dominated by low-wage informal sector
activity – and growth has not fuelled formal sector job
creation on the expected scale (USAID and DFID, 2014).
Productivity gains have been modest. Human capital
constraints have been identified as a growth bottleneck
and a source of skills gaps in the formal sector. There is
little evidence to suggest that more access to education has
reduced poverty through increased productivity (World
Bank, 2013).
Education has to be central to any strategy for
resolving these problems. For many Bangladeshi children,
education remains a Hobbesian ‘poor, nasty, brutish and
short’ experience followed by an early transition into the
world of work. Causation operates in both directions.
Limited access to poor quality education pushes many
children towards labour markets. Meanwhile, poverty
and vulnerability pulls many children into work even
when schooling options are available. Breaking the link
between child labour and education is recognised by the
Government of Bangladesh as a critical requirement for
achieving the 2030 goals.
Looking ahead to 2030, and the more ambitious
targets set under the SDGs, Bangladesh faces a number
of challenges. The current national strategy envisages
Child labour and education – a survey of slum settlements in Dhaka  19  
middle-income status by 2021, which will require a
steep increase in economic growth. The elimination of
extreme poverty is an achievable goal – but it will not be
achieved without more inclusive growth, human capital
development and expanded investment in infrastructure
(USAID and DFID, 2014). Similarly, sustained progress in
human development will require a greatly strengthened
focus on social groups and regions that have been left
behind.
2.1 Education for All – the unfinished
agenda
The headway that Bangladesh has made in education is
widely recognised (Bangladesh Bureau of Statistics and
UNICEF Bangladesh, 2014). There has been a dramatic
widening of access to primary and lower secondary
education with gender parity achieved at both levels.
Completion rates have improved and repetition rates have
fallen. Despite these gains, Bangladesh faces a ‘twin crisis’
in education – a crisis in access and learning. Millions
of the country’s poorest children do not complete a
full primary school cycle, let alone make the transition
to secondary school. Meanwhile, the school system is
characterised by low and highly unequal levels of learning.
The two sides of the crisis are linked, with poor quality
learning contributing to high drop-out rates.
2.1.1 Basic education from 6–14
Basic education in Bangladesh is structured across two
levels overseen by two separate ministries.8 Legislation
provides for free and compulsory education up to
Grade 5, with entry to Grade 1 nominally at 6 years of
age. With smooth progression across grades, primary
education would cover the ages 6–10. Junior secondary
schooling covers Grades 6–8 (or ages 11–13 in a normal
progression). Proposals in the 2010 National Education
Policy aimed at extending the primary cycle to eight years
have yet to be translated into policy decisions (Bangladesh
Ministry of Education, 2010). Secondary education covers
Grades 9–10. Junior secondary and secondary education
are neither compulsory nor free.
Schooling at the primary level involves a wide array
of providers. Government primary schools account
for around one third of providers, and one fifth are
recently nationalised primary schools brought under
government supervision. Madrashas9 account for around
7% of reported provision. There is also an extensive
array of non-formal provision dominated by BRAC and
8 The two separate ministries responsible for overseeing provision are the Ministry of Primary and Mass Education (Grades 1–5) and the Ministry of
Education (Grades 6–12).
9 Madrashas in Bangladesh are a non-graded education system that follows its own curriculum and is parallel to the mainstream education system. They
are religious schools or colleges where students are taught Arabic and the Islamic religion.
10 Upazilas are administrative sub-districts of Bangladesh.
20  ODI Report
Reaching-Out-of-School-Children (ROSC) centres. Most
junior secondary and secondary education is delivered
through non-government providers.
Bangladesh has a well-established track record in
expanding access through demand-side financing. All
children attending primary school are eligible for a small
monthly stipend (amounting to 100 Bangladeshi takas (Tk)
in 2013, about $1 at the time), which reached 7.8 million
children in 2013. The stipends are targeted at poor children
selected by school authorities. However, the value of the
stipend has eroded in real terms since its introduction
in 2002 – and households with two or more children in
school receive only 125% of the stipend (Government of
the People’s Republic of Bangladesh, 2013). The stipend is
applied at the same level across grades.
The junior secondary stipend for girls has contributed
to gender equity by creating an incentive for parents to
keep children in school. Girls making the transition to
secondary school in rural areas receive a stipend and
free tuition – and all children receive free textbooks. The
Female Secondary Stipend Programme, introduced in the
mid-1990s, provides a CCT to all secondary school female
students, combined with the transfer of a tuition fee to the
school they attended. Conditions are attached with regard
to school attendance and students remaining unmarried.
While causality is difficult to establish, data suggests
that the stipend programme has contributed to the rise
in enrolment of girls in secondary schools (Schurmann,
2009). Econometric analysis based on household data
indicates significant effects. On average, an additional
year of stipend programme duration increases female
student secondary enrolment by as much as 8%. The
household level data suggest that an additional year of the
programme increases enrolment of girls aged 11–18 by
12 percentage points (Khandker et al., 2013). Questions
remain as to the impact of the stipend programme on
delaying marriage, empowerment of girls and women and
enhancing employment opportunities.
More recently, the government has introduced a
poverty-targeted stipend programme (SEQAEP) in around
a quarter of Bangladesh’s upazilas.Students eligible for
the stipend receive from $15 to $40 a year, depending
on their grades, and benefits are conditional on students
maintaining 75% average attendance, achieving a passing
grade in final examinations and remaining unmarried until
they complete Grade 10. Means-tested stipends have been
issued to over 1.5 million students (Parajuli, 2016). Initial
evidence points to marked increases in enrolment among
the poor, with the programme contributing to a reported
20% increase in secondary school retention for males and
Figure 2. Bangladesh is nearing universal primary school
enrolment: reported enrolment rates between 2008–2015
15% for females, with larger effects for children from
by gender
poorer households (Jolliffe et al., 2013).

2.1.2 The country is nearing universal primary 120

enrolment – but retention is a concern
Bangladesh has dramatically improved access to education. 100
Expanded supply coupled with demand-side financing
through stipends has pushed the country towards universal 80
primary enrolment. The primary net enrolment rate was
most recently reported at 97% and the gross enrolment 60
rate (GER) at 109% (Figure 2). Between 2000–2010, the %
junior secondary GER rose from 52% to 62% in 2013 40
(Jolliffe et al., 2013). There are more girls than boys in
school at both levels. 20
Unfortunately, enrolment rates tell only part of the
story. Repetition rates have declined over time, but they 0
remain high (Figure 3). Around 7% of pupils repeat Grade 2008 2009 2010 2011 2012 2013 2014 2015
1, which points to widespread problems with school GER (%) Boys GER (%) Girls
readiness. One of the indicators for school readiness is GER (%) Total NER (%) Boys
attendance at pre-school in the year prior to entering NER (%) Girls NER (%) Total
primary school – and only 43% of children attend pre- Completion Total
school according to household survey data (Government
of the People’s Republic of Bangladesh, 2014).11 Repetition Data source: Bangladesh Bureau of Educational Information and
rates remain high in Grades 3–4. Statistics (BANBEIS), The Directorate of Primary Education (DPE)
Annual Primary School Census (APSC) 2015.

Figure 3. Repetition rates are falling – but remain high: repetition by grade and gender, 2011–2014

16
14.2
14 13.5

11.6
12 11.1
10.7 10.6
10.3
10 9.4
8.8
8.4
7.6 7.9 7.3 7.4 7.3 7.3 7.3
% 8 6.9 6.9
6.7 6.5

4 3.5
2.1
1.7
2

0
1 2 3 4 5 Boys Girls Total
Grade Gender

2011 2012 2013

Data source: Government of the People’s Republic of Bangladesh (2014) Bangladesh Primary Education Annual Sector Performance Report
– 2014. Dhaka: Monitoring and Evaluation Division Directorate of Primary Education.

11 The Directorate of Primary Education reports a higher figure of 65% for 2013.

Child labour and education – a survey of slum settlements in Dhaka  21  

 Many of the children entering primary education will
not complete a full cycle. In 2013 the drop-out rate across
the primary cycle was 21%, with a peak at Grade 4 (Figure
2.3). Drop-out rates for girls have fallen far more rapidly
than those for boys, expanding a reverse gender gap.
The upshot of these patterns is that of every 10 students
entering Grade 1 only around 7–8 progress to Grade 5,
6–7 transit to Grade 6, and 4–5 complete a secondary
cycle (BANBEIS and Bangladesh Ministry of Education,
2015). Across the full secondary cycle, girls have higher
drop-out and lower completion rates than boys from
Grade 6. Longitudinal evidence has identified a number
of risk factors for early drop-out (Sabates et al., 2010).
Grade repetition and late entry into primary school figure
prominently. Poverty, low levels of parental education and
time spent on household chores are all associated with
early drop-out.
Figure 4. Many children start school but do not complete
a full primary cycle: drop-out rates by grade and gender,
2015
25
20
15
% The risk of being out of school is not equally distributed
10
5 being born into a poor household is a major risk factor.
0 2013 compared to 89.4% for the middle wealth quintile
Grade 1
Grade 2
Grade 3
Grade 4
Grade 5
Grade 6
Primary Education
Boys Girls
Data source: BANBEIS and Bangladesh Ministry of Education (2015)
Bangladesh Education Statistics – 2015. Dhaka: Bangladesh Bureau
of Educational Information and Statistics (BANBEIS).
Drop-out patterns have an important bearing on the
link between education and work. As we show in Section
3, from the age of 10 in the case of boys and 11 for girls,
children in urban slums face elevated risks of a premature
transition from school to work. The ability of children
to secure wages outside the home increases beyond
these thresholds, as do the associated opportunity costs
of keeping children in school. With a combination of a
one-year delay in entry to school and repetition of an early
grade, children would reach the age of 10–11 at Grade
3–4 well before completing a full primary cycle. Given
22  ODI Report
that net intake data for Grade 1 (children attending at the
stipulated age of six) from the Multiple Indicator Cluster
Survey (MICS) indicates that only around one third of
children are attending at the right age-for-grade, delayed
entry to school is a risk factor for working children
(Bangladesh Bureau of Statistics and UNICEF Bangladesh,
2014).
Gaps between reported enrolment and the school
attendance levels recorded in surveys are indicative of high
drop-out rates. School enrolment figures in Bangladesh
are derived from administrative reporting by schools and
education authorities. They are notoriously unreliable,
in part because school funding is linked to enrolment.
Education authorities have acknowledged a strong bias
towards overestimation of enrolment and underestimation
of over-age children (Bangladesh Bureau of Statistics
and UNICEF Bangladesh, 2014). Survey evidence on
school attendance by primary school age children paints
a less encouraging picture. Figure 5 compares national
administrative data on school enrolment with data from
the 2014 Demographic and Health Survey. It is worth
noting that a 2012/2013 MICS survey reported even lower
levels of net attendance, at just 71%. School participation
falls sharply at secondary school level, albeit with a
narrower gap between enrolment and attendance.
2.1.3 Social disparities limit school participation
across social groups – and Bangladesh has some distinctive
patterns, as illustrated in Figure 5. As in other countries,
Net attendance rates for the poorest 20% were 84.4% in
Grade 7
Grade 8
(Bangladesh Bureau of Statistics and UNICEF Bangladesh,
2014). In marked contrast to the pattern in other low-
income countries, school participation in Bangladesh is
Junior not associated with an urban advantage – and indicators
Secondary for boys lag behind those for girls. Data from the 2014
Education
Demographic Health Survey highlights the precarious
All position of urban boys and, more broadly, boys from the
poorest homes (NIPORT et al., 2016). There are some
significant differences across national divisions, with
Sylhet, Chittagong and Dhaka exhibiting lower levels of
attendance and enrolment than divisions in the west of the
country (World Bank, 2013; UNESCO World Inequality
Database on Education (WIDE) – Data on Bangladesh,
2014). The ‘gender advantage’ in attendance enjoyed by
girls in primary education is more limited for secondary
attendance, with the poorest boys and girls falling far
behind.
While precise figures are difficult to estimate, probably
around 5.6 million children aged 6–13 are out of school.
Some of these children will be late entrants (UNICEF,
2014). There is a small minority who will never enrol.
However, the majority will have enrolled and dropped out
before making the transition to junior secondary school.
 Government strategies recognise the need to extend the
Table 1. Primary gross and net attendance rates: slums
versus the urban average, 2011
reach of the primary school system. The ROSC project,
which is supported by the World Bank, explicitly targets
hard-to-reach children in 148 remote and disadvantaged Gross attendance rate Net attendance rate
upazilas, providing a combination of student grants, books
Slum 91 62
and teacher support. An estimated 690,000 children have
reportedly been reached through contacts with learning Boys 86 59
centres (Al-Zayed, 2016).12 However, student retention Girls 96 66
in these centres remains problematic and the project has
Urban average 102 77
been introduced only on an initial pilot basis in urban slum
areas. Data source: Government of the People’s Republic of Bangladesh
(2014) Bangladesh Primary Education Annual Sector Performance
Figure 5. Attendance lags behind enrolment, with marked Report – 2014. Dhaka: Monitoring and Evaluation Division
inequalities: enrolment and attendance rates, selected Directorate of Primary Education.
groups in 2013
Supply-side constraints figure prominently for slum
97.0
100 populations. There were just 564 government primary
86.4 → Female, Rural 88% schools located in slum areas in 2013 – half of them
90 → Female, Poorest 86.9%
→ Male, Urban 84.2%
80 → Male, Poorest 81.9% in Dhaka (Government of the People’s Republic of
Bangladesh, 2014). The World Bank survey found there
70
was one primary school near the entrance of a slum for
60 every 121 children, and one secondary school for every
48.2
50 42.8 → Female, Rural 44% 678 children. It should be emphasised that these schools
40 → Male, Urban 40.7% serve both slum and non-slum communities – and that the
30 → Female, Poorest 30.8% vast majority of secondary schooling is delivered through
→ Male, Poorest 28.2%
private providers.
20
The experience of slum-dwelling children serves
10 to illustrate some of the wider pressures facing poor
0 households. Although education at the primary level is
Net primary Net primary Net Net secondary
school school secondary school nominally free, out-of-pocket payments play an important
enrollment attendance school attendance role in education financing. The monthly education
rate (2013) rate enrollment rate
(DHS, 2014) rate (2013) (DHS, 2014) expenditure reported for a primary school student in 2011
was Tk 370 (or 3.7 times the primary school stipend)
Data source: Bangladesh Bureau of Statistics, 2013; Demographic
(World Bank, 2013). By far the largest share of spending
Health Survey, 2014.
reported by slum dwellers is directed to private tuition,
Children living in urban slums figure with some with non-poor students spending twice as much as poor
prominence in Bangladesh’s out-of-school population. On students. To the extent that private tuition spending
average, these children participate far less in education reduces the risk of repetition and drop-out, it may have the
than their non-slum urban counterparts and children in effect of exacerbating disparities in school participation.
rural areas. Gross and net attendance rates for children
aged 6–10 are well below the national average (Table 1). 2.1.4 Progress in learning has lagged behind pro-
Research by the World Bank based on the 2011 Urban gress in access
Slum Survey and the 2010 Household Income and Bangladesh’s success in expanding access to education
Expenditure Survey found high rates of enrolment (in the stands in stark contrast to the record on learning.
range of 80–90%) for slum dwellers at 8–9 years old, with Education achievement levels are low and highly unequal.
far lower rates (in the range 50–60%) at 6–7 years old – an For many children, the marginal learning effect of an
indicator for late enrolment. However, enrolment rates additional year in school is close to zero. Social differences
fell steeply from the age of nine to just 60% by the age of in learning outcomes are evident in early grades and
11. Within this picture of overall disadvantage, there are continue across the primary and secondary school cycles.
marked differences between the poor and the ‘non-poor’ Student-based learning assessments underscore the
living in slum areas. weak link between school participation and learning. The
2013 National Learning Assessment (NLA) ranks students

12 ROSC centres are not strictly primary education facilities. Students tend to be older (between 8 and 14 years of age) than regular primary school students,
and students and teachers follow a flexible school timing to suit their mutual needs. Students are taught by a single class teacher, until they are ready to
appear for the Grade 5 examination and can then join the mainstream secondary schools.

Child labour and education – a survey of slum settlements in Dhaka  23  

on a competency scale from 1–5 for Bangla and maths
at Grades 3 and 5, with each band approximating to the
learning level expected for the grade. Around one quarter
of Grade 3 students performed at the lowest band for
Bangla. For maths the figure rises to 43%. While there is
a marked improvement by Grade 5, this may owe less to
learning gains than to high drop-out rates among students
at the lower end of the learning distribution. In the case of
maths, just one quarter of Grade 5 students are learning at
the level expected and 40% are learning at Grade 3 or less.
Breaking down these learning achievement outcomes
highlights the interaction of school-based and home-based
inequalities. The NLA estimated that differences between
schools accounted for around three quarters of variations
in score and student-related characteristics for one quarter.
Government primary schools performed slightly better
than other schools, though these schools play a limited
role in serving the most disadvantaged students in slums
(MoPME, 2014).
Many children that make the transition from school
emerge from school with limited levels of learning. One
recent survey used a representative sample of over 3,000
10–17 year-old rural children to assess ability on basic
arithmetic. Of children who had completed primary school,
just 52% of males and 42% of females could answer
three or more of four questions correctly (Asadullah and
Chaudhury, 2013). Findings such as these raise serious
concerns over the alignment between the quality of
education delivered in Bangladesh’s schools and the type
of education needed to drive poverty reduction, inclusive
economic growth and human development.

Figures 6a and 6b. Learning deficits remain large: learning achievement at Grades 3 and 5 for Bangla (Figure 6a) and
Maths (6b), 2013

Grade 3 5% 20% 40% 27% 8%

Grade 5 3% 20% 52% 25%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Band 1 Band 2 Band 3 Band 4 Band 5

Grade 3 15% 28% 33% 20% 4%

Grade 5 11% 10% 30% 34% 25%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Band 1 Band 2 Band 3 Band 4 Band 5

Data source: Government of the People’s Republic of Bangladesh (2014) Bangladesh Primary Education Annual Sector Performance Report –
2014. Dhaka: Monitoring and Evaluation Division Directorate of Primary Education.
Legend
Band 1: Students working well below Grade 3 level
Band 2: Students working below Grade 3 level
Band 3: Students working at Grade 3 level
Band 4: Students working above Grade 3 level
Band 5: Students working at Grade 5 level

24  ODI Report

2.1.5 Bangladesh’s twin crisis – some underlying
drivers
The underlying sources of the twin crisis in access and
learning in Bangladesh have been extensively explored.
Underinvestment in school infrastructure is part of the
problem. Government funding for education represents
2.1% of gross domestic product (GDP), which is low by
international standards (it averages 5% of GDP in sub-
Saharan Africa). Reporting from teachers suggests that
over two thirds of classrooms in government schools are in
‘poor’ or ‘very poor’ conditions.
Many children experience classroom conditions which
are not conducive to effective learning. While around
86% of schools report a student-to-teacher ratio below 46
(which is higher than best practice international norms),
which includes second-shift schools. Children in Grades
1–2 in these schools attend school for only 520 hours a
year on average compared to an international standard of
900–1,000 hours (Government of the People’s Republic of
Bangladesh, 2014). Teacher absenteeism is also widespread.
It is the most disadvantaged students who bear the
brunt of infrastructure deficits. The poorest quintile of
students face the highest student-to-teacher ratios (58).
They are also taught by teachers with fewer qualifications.
Just one fifth of schools in the poorest upazilas have access
to electricity. These infrastructure disadvantages appear to
be reinforced by capital spending which, unlike recurrent
spending, is skewed towards wealthier upazilas (Steer et
al., 2014).
Cutting across all of these issues are concerns over
teaching practices. Government teachers at the primary
level receive minimal training in approaches to cognitive
development. The training they do receive places an
emphasis on rote learning and memory recall. For children
who have no experience of pre-school, come from homes
marked by limited literacy, and enter school with low
expectations, these are particularly damaging teaching
models.
2.2 Child work and child labour
Child labour constitutes an obstacle to Bangladesh
achieving the 2030 development goals. It harms the well-
being of individual children, curtails their opportunities
and locks them into a cycle of disadvantage. Children
forced out of school and into work either by economic
circumstances or failures of the education system are
denied the chance to develop the skills and competencies
they need to realise their potential, find secure and
remunerative employment and break the transmission
of poverty across generations. The loss of educational
opportunity for these children also represents a source of
capability deprivation that limits their choices and freedom
(Sen, 1999). Beyond the individual cost, Bangladesh cannot
afford to squander the human capital assets of its young
population if the country is to secure a demographic
dividend.
2.2.1 Child labour is widespread, but the numbers
depend on definitions
Child work and child labour are widespread in Bangladesh.
There are uncertainties about the extent of both practices –
and the boundaries are blurred. Following the Convention
outlined in Section 1, the 2013 Child Labour Survey
(CLS) published by the Bangladesh Bureau of Statistics
(BBS) differentiates between different forms of child
work, and between hazardous and non-hazardous child
labour (Bangladesh Bureau of Statistics, 2013). Using a
representative household survey, the CLS estimated that
3.4 million of Bangladesh’s 5–17-year-old children were
working, or around 4% of the age group. Just under half
of these children – 1.6 million – were working under
conditions that constitute child labour, three quarters of
them in hazardous labour (Figure 7).
Figure 7. Estimated child work and child labour in
Bangladesh, 2013
Population aged
5–17 years
39,652,384
Working children
3,450,369
Child labour
1,698,894
Hazardous child
labour
1,280,195
Data source: Bangladesh Bureau of Statistics (2013) Child Labour
Survey Bangladesh 2013. Dhaka: Bangladesh Bureau of Statistics and
International Labour Organization.
These figures have to be interpreted in the light of
Bangladesh’s distinctive legislation. The Labour Act
of 2006 defines 14 as the minimum age of entry to
employment – and as the threshold separating a ‘child’
from an ‘adolescent’. Children aged 5–11 are not permitted
to enter employment. However, from the age of 12,
children are permitted to carry out ‘light work’ for up
to 42 hours (the ILO reference standard is 14 hours),
provided it does not interfere with their education or
affect their health. Children and adolescents aged 5–17 are
prohibited from working more than 42 hours. Employment
in excess of this level is defined as ‘hazardous labour’. In
2013 the government also identified 38 types of work as
hazardous activity constituting ‘the worst forms of child
Child labour and education – a survey of slum settlements in Dhaka  25  
labour’. However, this list is more restrictive than the ILO’s
list.
It follows that the CLS survey is not directly comparable
in terms of what is being measured to the ILO global
survey. Based on data from a 2006 labour force survey
and a household survey from the same year, analysis by
Understanding Children’s Work (UCW) estimated the
number of children labouring in activities falling under the
‘for elimination’ umbrella indicated by ILO Conventions
at 5.1 million – including 3 million in the 5–14 age range
(UCW, 2011). Measured on a broadly comparable scale,
the incidence of child labour in Bangladesh was among the
highest in South Asia (Khan and Lyon, 2015).
Child labour in Bangladesh spans a vast range of
activities. One of the distinctive characteristics of
Bangladesh compared to other low-income countries is
the high level of reported child labour in manufacturing.
According to the CLS data, manufacturing accounts
for around one third of the total (and a slightly higher
share of hazardous labour), or slightly more than the
share in agriculture and forestry. Services such as retail,
wholesale and transport also figure with some prominence,
accounting for another 15%. Child domestic labour is not
covered in the 2013 survey, but it is a common form of
employment. An ILO study put the number involved at
over 400,00 in 2006, with children from poor households
in rural areas providing the main source of labour supply
(ILO, 2006).
Some activities are strongly associated with child labour.
Significant numbers of children are employed as ‘bidi
workers’13 in the cigarette industry, especially in northern
areas, and in the dried fish industry in Chittagong and
Kuakata (UCW, 2011). As is evident from even casual
observation in urban areas, many children are involved
in portering activities, street vending and transport-
related activities, ranging from ticketing for buses to
rickshaw pulling and cleaning vehicles. More hidden is
the employment documented in hazardous sectors such
as welding, metallic workshops and machining. The
large number of street children in the country eke out an
existence hawking, collecting and recycling papers and
garbage, polishing shoes and other activities.
The CLS provides a valuable snapshot of the national
child labour profile, as indicated by national legislation.
That snapshot makes it possible to disaggregate reported
child labour by school age group. Survey estimates put
the number of child labourers in the primary education
age cohort of 6–11 at around 430,000, with another
39,000 aged 12–13, and 1.2 million aged 14–17. Dhaka
and Chittagong divisions dominated reported child labour
numbers, accounting for over one million of the national
total. City corporations accounted for just 250,000
labourers. More boys were reported to be working than
girls.

Figure 8. National surveys point to an overlap between school and work: education status by age of children involved in
work, 2013

100
87.2 87.2
90
80 73.1

70 63.0
61.1
55.3 55.3 55.4
60 54.5

% 50 40.5
40 34.7 34.7
30.9
28.6
30 21.9 22.7
19.2
20 10.0 10.0 9.8 9.8
8.0 8.4 7.7
10 3.1 3.1 5.0

0
Currently Currently Never Currently Currently Never Currently Currently Never
attending not attended attending not attended attending not attended
attending s chool attending s chool attending s chool
Children aged 6 to 11 Children aged 12 and 13 Total

Working children Child labour Hazardous labour

Data source: Bangladesh Bureau of Statistics (2013) Child Labour Survey Bangladesh 2013. Dhaka: Bangladesh Bureau of Statistics and
International Labour Organization.

13 Bidis are locally-produced cigarettes.

26  ODI Report

 The 2013 survey found that many working children
and child labourers were also attending school (Figure
8). An estimated 29% of child labourers were in school,
declining to 19% for children involved in hazardous
work. The earlier UCW survey also found a large overlap
between employment and child labour, with just under 7%
of 7–14-year-olds engaged in both activities and 4% just
employment (UCW, 2011).14
Comparisons with earlier estimates suggests that child
labour levels are falling, but this has to be treated with
caution. Estimates based on employment survey data for
2005 put the number of children working at 3.6 million
for the 7–14 age group (UCW, 2011). Around two thirds
of these children were also attending school. The same
survey provided an estimate for child labour among the
5–17 age group (comparable to the 2013 CLS) at 5.1
million. However, this used a ‘child labour for elimination’
international benchmark which is only partially aligned
with the definition of child labour used by authorities in
Bangladesh.
The 2013 CLS provides detailed information on hours
of work. On average, working children were found to be
working for 39 hours a week. Two thirds of child labourers
were working more than 42 hours a week, rising to 82%
and 92% respectively for male and female workers in
hazardous sectors. Children working in City Corporation
areas were working the longest hours (Bangladesh Bureau
of Statistics, 2013).
2.2.2 The national environment
Bangladesh has adopted a wide range of policies on child
labour. The overall framework defined in legal terms by the
2006 Labour Act (Government of the People’s Republic
of Bangladesh, 2006) and its subsequent amendments is
underpinned by the National Child Labour Elimination
Policy (2012–2016) (Ministry of Labour and Employment,
2013), which includes a commitment to eliminate the
worst forms of child labour by 2016, and the National
Children Policy. The national strategy for eliminating child
labour involves a wide range of ministries, programmes
and international partners.
While the Government of Bangladesh has made strong
commitments to eradicate child labour, national legislation
falls short of the standards set in ILO Conventions. The
country has not ratified ILO Convention 138 on the
minimum working age, though it has ratified Convention
182 on the worst forms of child labour.
Important gaps remain in terms of children’s legal
protection (UCW, 2011). Beyond the garment sector, efforts
to enforce child labour prohibitions are limited. During
2013, the Department of Inspection for Factories and
Establishments filed just six cases of violations with the
14 Note that the UCW definition of employment is not equivalent to the CLS definition of child labour.
Bangladesh Child Labour Court (US Department of Labor,
2013). Fines for violations are set at levels insufficient
to have a deterrent effect. Unannounced site visits are
infrequent, especially in unregistered establishments. Our
survey of child labour in Dhaka’s slums suggests that the
garment sector itself may be regulated far less effectively
than has been assumed (see Section 3).
Beyond the weak regulatory capacity, policies suffer
from a lack of coherence. While the National Education
Policy sets clear goals, it does not provide a comprehensive
strategy for enforcing compulsory education provision,
keeping vulnerable children in school, or attracting child
workers and other out-of-school children back into school.
More broadly, the country lacks an integrated strategy that
combines social protection, education, employment and
wider interventions to eliminate child labour.
Our survey calls into question some aspects of the
picture to emerge from the CLS of 2013 and other surveys.
Five core issues, examined in more detail in Section 3,
stand out:
Scale of the problem: Based on a large-scale
representative study of Dhaka slum settlements with a
population in excess of 400,000, we find an incidence of
child labour – around 15% – higher than might have been
expected on the basis of the CLS.
The blurred boundary between child work and child
labour: Unlike previous surveys, including those by the
CLS and the UCW, we do not find a marked division
between child work and child labour, either as defined
by ILO standards or by Bangladeshi legislation. The
overwhelming bulk of child work in Dhaka’s slums appears
to constitute child labour, with a marked skewing towards
hazardous labour. This points to a very different pattern of
child work in slums in comparison with rural areas.
Working children in slums are overwhelmingly out-of-
school: Earlier surveys have found a large share of working
children combining school with work. By contrast, we find
a high level of segmentation – few children are combining
work with school. If accurate, this is a finding with
important policy implications, since it implies that entry
into the world of work implies exit from education.
There is a steep age gradient with labour market
entry taking off from the age of 10. Partly because of
age-ordering, the 2013 CLS finds a low incidence of
child labour among children aged 6–11, with a far higher
incidence in the 14–17 cohort. Our findings broadly
endorse the implied age gradient. However, the Dhaka
slum survey finds high levels of entry into the world of
work at age 10 for boys and 11 for girls, and by age 14
almost half of children are working.
There are high levels of employment in the formal
garment sector. Contrary to our expectations, we found
Child labour and education – a survey of slum settlements in Dhaka  27  
children reporting high levels of child labour in the formal
garment sector. For girls, the sector is the largest employer.
However, in the absence of detailed enterprise information
these findings have to be treated with caution.
Most child workers in slums exceed the 42 hour
working limit marking the threshold under national
legislation for child labour: Our slum survey did not reveal
a clear distinction between child work and child labour
as defined by the 42 hours a week threshold. The average
working week reported by children in our sample was 64
hours, rising to 70 hours for the median child worker. Only
15.5% reported working for less than 42 hours a week.
2.3 Dhaka’s urban slums – the wider
environment
Slums provide a highly visible reminder of the inequality
that has accompanied Bangladesh’s development. Many
have grown in close proximity to the enclaves of wealth
and prosperity that house Dhaka’s wealthier residents.
Badda slum, one of the sites covered in our survey, borders
two Dhaka’s most affluent areas – Gulshan and Baridhara
– and is home to a population estimated to be in excess
of 100,000. Kamrangir Char, which is located on what is
effectively a flood plain of the Buriganga River, is similarly
located adjacent to prosperous areas in Dhaka South.
Dhaka is the primary engine of urbanisation in
Bangladesh, and one of the world’s fastest-growing
megacities. In the decade after 2000, the population
of Dhaka Metropolitan Area expanded by around five
million (Cox, 2012). Around one third of the country’s
urban population is now concentrated in these areas. In
total, there were around 15 million urban inhabitants in
Dhaka division in 2012, including up to half-a-million new
migrants that arrive each year (Gruebner et al., 2014). The
vast majority of these migrants are from rural areas. Much
of the expansion of settlements in Dhaka has occurred
in slums. In addition to absorbing an inflow of rural
populations, there is a high level of mobility across slum
areas in the city.
None of the world’s other megacities comes close to
matching Dhaka’s population density. At over 45,000
people per square kilometre, Dhaka has a density level
one third greater than Mumbai, which stands second in
the megacity population density league table (Cox, 2012).
On one analysis of demography, migration and economic
trends carried out by McKinsey, Dhaka is projected to
become the world’s seventh most populous city by 2025.
Of considerable relevance in the context of the current
study, the same projection suggests that Dhaka could hold
15 The Bangladesh Bureau of Statistics Census surveys identifies six characteristics of a slum area: small dwellings built of cheap materials, high population
density, a concentration on government and semi-government land, inadequate water and sanitation, inadequate lighting and road facilities, low
socioeconomic status (Bangladesh Bureau of Statistics, 2014).
28  ODI Report
the world’s third largest population of children (Dobbs et
al., 2011).
Dhaka’s slums have grown with the growth of the city’s
population and the economy. Slum settlements are highly
dispersed. There are marked concentrations around areas
with significant manufacturing and industrial units in
Dhaka North, as well as port transit points, markets and
hubs of informal sector activity.
There is no clearly delineated definition of a slum areas
in Bangladesh.15 This has led to widely varying estimates of
the size of Dhaka’s slum population. According to the 2014
Census of Slum Areas, there were 6,489 slums in Dhaka
Division, housing around one million people – or just
under half of the national total both by number of slums
and slum households. These slums were spread across four
city corporations. In 2011, the old DCC was divided into
the two separate administrative entities of Dhaka North
and Dhaka South, which account for around one quarter
of all slums in Bangladesh. The 2014 Slum Census provides
estimates for the number of slums and slum populations
(Table 2). Measured by population, Dhaka North accounts
for, by far, the largest slum population in the entire country.
The 2014 census put the figure at just under 500,000,
compared to 147,000 in Dhaka South. The third City
Corporation, Gazipur, accounts for another 8% of the
slums and a population of 185,000 recorded in the 2014
census. Narayanganj, the fourth Corporation, houses a
small number of slums.
The real size of Dhaka’s slum population is uncertain.
Survey-based estimates of the slum population have ranged
from around one third to over 60% (Cameron, 2009).
The Centre for Urban Studies estimated in 2006 that there
were over 4,000 slums in Dhaka, with a bastee (slum
and squatter settlement) population of 3.6 million – far
larger than the estimate in the 2014 Census (Islam et al.,
2006). These differences reflect well-known difficulties
in estimating slum populations. Census estimates often
exclude unregistered dwellings and slum dwellers, and the
dividing line between formal and informal settlements is
often blurred (Lucci and Bhatkal, 2014). Aerial surveillance
systems can also under-represent slum households,
especially in areas marked by high population densities on
sloping areas (Lucci and Bhatkal, 2014).
While the term ‘slum’ captures some shared
characteristics, it also obscures important differences. Slum
dwellers in Dhaka are overwhelmingly poor, but there
are marked differences in levels of deprivation within and
across slums (Angeles et al., 2009). These differences are
reflected in occupational structures, income, assets, the
quality of shelter and whether or not slum dwellers own or
rent their properties (Cameron, 2010; 2011). As we show
below, inequalities within the slum-dwelling community
have significant implications for patterns of children’s
work and education.
Monetary poverty is a pervasive feature of slum life.
Poverty in Dhaka has been estimated at 15% for extreme
food poverty and 31% for a higher threshold (Ahmed et
al., 2010). More detailed poverty mapping reports poverty
levels in excess of 55% for the poorest 10% of upazilas in
Dhaka (Ahmed et al., 2010: 7). While poverty assessments
do not allow for reliable estimates of poverty levels and
trends for slums, they indicate poverty levels in Dhaka
stagnated over the period 2005-2010 despite the marked
national decline. This represented a reversal of the pattern
observed between 2000 and 2005 (Jolliffe et al., 2013).
Migration from the poorer and more rural Barisal and
Rajshahi divisions to the urban Dhaka division could have
been a contributory factor.
Social conditions in Dhaka’s slums reflect the unplanned
nature of the urbanisation process. Surveys carried out by
the CUS and others provide valuable snapshots of slum
life. Population density levels are exceptionally high: some
informal settlements have densities in excess of 4,000
people for every acre, which translates into the equivalent
of more than 1 million per kilometre (Cox, 2012). Over
90% of dwellings in slums comprise just one room less
than 14 square metres in size (Ahmed, 2014). Many slums
are located in low-lying areas near rivers that are prone to
flooding (Mohit, 2012).
Even the most basic infrastructure is often lacking. Most
slum dwellers have access to municipal water supplies
through stand-pipes. However, sanitation and drainage is
limited. The CUS survey found that only 10% of slums
had adequate drainage and that one third of slum dwellers
lacked access to safe sanitation. Larger slums have greatly
outgrown the minimal sanitation infrastructure that is
available. Only 30% of the population in slums with over
500 people covered in the CUS survey had access to safe
sanitation.16 Another survey of Karail, among the largest
slums in Dhaka, found around half of households had
access to safe sanitation (Islam et al., 2015).
Limited access to safe sanitation and poor drainage
expose slum-dwelling populations to acute health risks.
Flooding and water-logging are frequent occurrences.
Coupled with poverty and high levels of population
density, this creates fertile conditions for the transmission
of communicable diseases among children, while the
limited provision of quality public health services hampers
effective treatment and prevention. Poor nutrition is both
a cause and consequence of elevated health risks. One
survey in Kamrangir Char slum, conducted in 2011, found
that 4% of children reporting to health clinics showed
signs of severe acute malnutrition and 23% severe chronic
malnutrition. Prevalence was highest among children aged
13–24 months.

Table 2. Dhaka City Corporation slum areas and population

Percentage of the total

Percentage of the total
Number of households number of households Total number of
Dhaka Division Number of slums slum population in
living in the slums living in slums in urban slum dwellers
Bangladesh
areas of Bangladesh
Dhaka North City 1,644 135,061 22.8 496,669 22.3
Corporation
Dhaka South City 1,755 40,015 6.7 147,066 6.6
Corporation
Gazipur City Corporation 1,286 56,597 9.5 185,867 8.3
Narayangonj City 81 10,804 1.8 40,485 1.8
Corporation
Total in Dhaka Division 6,489 292,780 49.4 1,061,699 47.7
Total in Bangladesh 13,938 592,998 100.0 2,227,754 100.0
urban areas

Data source: Bangladesh Bureau of Statistics (2015) Preliminary Report on Census of Slum Areas and Floating Population 2014. Dhaka:
Bangladesh Bureau of Statistics

16 Defined in this case as access to sealed latrines or latrines linked to sewers or septic tanks.

Child labour and education – a survey of slum settlements in Dhaka  29  

 Slum dwellers face some distinctive patterns of
vulnerability. While most slums are located on government-
owned or private land, informal landlords often collect
rent and can charge exorbitant fees for basic services. What
has been termed the ‘informal structure of extortion and
crime’ is widespread. Security of tenure is limited, with
almost 80% of slum dwellers paying rent (Islam et al.,
2006). Even where poor households enjoy legal ownership,
their ability to enforce claims and entitlements is limited
(Shafi and Payne, 2007). The resulting insecurity exposes
slum dwellers to the constant threat of eviction and deters
investment by slum dwellers.
Children living in slums across Bangladesh face
restricted opportunities for education – and those in
Dhaka’s slums are no exception. The Directorate for
30  ODI Report
Primary Education puts the net school attendance rates
for children aged 6–10 in slums at 62% (Government of
the People’s Republic of Bangladesh, 2014). Government
provision is limited. There are fewer than 300 government
primary schools in Dhaka’s slums. While the situation
may have improved since 2007, when it was estimated
only around one quarter of slums had a government
school (Baker, 2007), provision remains limited. Non-
governmental organisations have partially filled the
resulting gap, along with a wide range of private and
religious providers. The patchwork of providers is reflected
in uneven levels of provision: one survey of four slums in
2008 found enrolment rates ranging from 56% to 83%
(Cameron, 2009).
3. Child labour and
education in Dhaka’s
slums
The Child Work and Education Survey (CWES) was
designed to provide a window into the world of work and
school as experienced by children living in Dhaka’s slums.
It was developed and implemented by BRAC Institute
of Educational Development, BRAC University (BIED,
BRACU) and the Overseas Development Institute.
The survey is not representative for slums in Dhaka,
and we caution against drawing generalised conclusions
for other urban slum areas. Interactions between education
and children’s work are shaped by the social characteristics
of slum populations and local labour market conditions.
However, our study combines a large-scale survey of
children aged 6–14, with insights from detailed focus
group discussions. The survey covers some of the most
populous and fastest-growing slum settlements in
Bangladesh and, by extension, the world – and the results
highlight a number of concerns over the relationship
between urbanisation, education and child labour.
In this section we outline the methodology used in the
survey, summarise the main findings and present the results
of an econometric study exploring the social characteristics
of children who are working rather than attending school.
3.1 The Child Work and Education Survey
(CWES)
The Dhaka City Corporation CWES is a unique exercise.
To the best of our knowledge, it is the largest survey of
its type carried out in Dhaka – or any other urban area in
Bangladesh. The survey was designed to provide a window
on the world of child work and education among children
living in some of Dhaka’s largest and fastest-growing
slum communities. The CWES was implemented in eight
Thanas spread across the City Corporation areas of Dhaka
North and Dhaka South. Within these areas we selected
Thanas and sub-Thana enumeration areas on the basis
of a set of deprivation indicators associated with slums.
In the absence of detailed census data, we carried out
our own household listing, collecting data on over 4,000
households. From this, we identified 2,700 households
with children aged 6–14 and conducted a questionnaire-
based survey with them and their parents. Part of the
value of the survey is that it provides a child’s-eye view of
the relationship between school and work on the basis of
randomised and representative sampling (Box 1).

Box 1. The CWES in context

Other surveys have reviewed the extent of child work and its interaction with education from different and more
partial perspectives. Our survey covers fewer households in Dhaka than the nationally representative 2013 Child
Labour Survey and is less representative of Dhaka Division as a whole. However, the 2013 exercise makes no
attempt to create representative samples for slums.a Another Urban Slum Survey carried out by the World Bank
used a stratified two-stage random sampling methodology to produce a representative sample of slums, drawing
on a 2005 database. This exercise covered 2,464 households across 50 slums in Bangladesh, with 30 located
in Dhaka.b The CWES covers a larger number of households for eight slums just in DCC areas based on 2015
listings. Apart from responding better to the marked increase in slum population size, our sampling is designed to
capture the extent and character of child work.
a The 2013 Survey covers 138 non-rural Primary Sampling Units/Enumeration Areas in Dhaka, 65 of them in City Corporation areas.
b For purposes of comparison, an Urban Slum Survey conducted by the World Bank draws on a 2005 database developed by the Centre for Urban
Studies. It includes 2,464 slum households drawn from 30 slums in Dhaka, 10 slums in Chittagong and 5 slums in both Khulna and Rangpur.
Our survey covers a larger group of households in a smaller number of slums.

17 Technical advice on the survey was provided by UCW, along with detailed comments of the design of the questionnaire, interpretation of results and the
econometric exercise summarised in this report.

Child labour and education – a survey of slum settlements in Dhaka  31  

Figure 9. Thana survey locations

Badda Thana
Shah Ali (part)
Kafrul Badda Thana
Thana Thana
Shah Ali (part)
Kafrul
Thana Thana
Badda Thana
(part)
Badda Thana
Adabor Tejgaon Ind. (part)
Thana Area Thana
AdaborKhilgaon Tejgaon Ind.
Thana Thana Area Thana
Khilgaon
Thana

Kamrangir Lalbagh Thana

Char Thana
Kamrangir Lalbagh Thana
Char Thana

Bangladesh

Dhaka

Slum area by ward/area in 2010

Slum area by ward/area in 2010

 The methodology deployed for the survey is explained
in detail in Annex 1. Briefly summarised, the design and
implementation phase involved four steps:
Step 1: Thana selection. Dhaka City Corporation area
spans 28 Thanas. Our concern was to identify a spread
of survey sites in Dhaka North and Dhaka South. We
selected Thanas through a partial ordering process by
ranking each Thana on the basis of proxy slum indicator
criteria, including population density, type of shelter, access
to safe sanitation, illiteracy and tenure security. Data was
drawn from the 2011 Population and Housing Census.
We identified eight Thanas as being among the worst
performing on at least three of the five criteria. Five of
these Thanas were in Dhaka North and three in Dhaka
South (Figure 9). Annex 1 includes detailed data on the
ranking of Thanas by our selection criteria.
Step 2: Survey site selection. We applied another layer of
selection criteria for the eight Thanas aimed at identifying
wards, units, parts and sub-parts containing slum
locations for survey purposes. Instead of ranking units,
we introduced four cut-off thresholds (again taking into
consideration budget and time availability) as follows:
•• Over half of households living in poor housing
•• Over half of households with either poor or no
sanitation
•• Over half of households renting
•• Illiteracy rates in excess of 25%
Two additional criteria were applied. First, for sampling
purposes we introduced a threshold requirement that
at least 150 households be present. Second, because our
specific concern was to understand the interaction between
child labour and educational disadvantage, we introduced
a requirement that at least 30% of children aged 6–14 in
Figure 10. Share of population lacking access to safe sanitation in Bangladesh’s Thanas
80
70 60
% of households
60 50
50
40
30
20
10
0
the administrative units to be not attending school. This
led to the identification of 38 administrative units across
the eight Thanas that met all the criteria. Conditions
within these units varied, reflecting the differences between
slum locations across Dhaka. Figures 10 and 11 illustrate
the diversity of the Thana slum environments for two of
our selection criteria – unsafe sanitation and reported adult
illiteracy – which have a marked bearing on education
prospects for children.
Step 3: Sample selection. Having identified potential
locations for the survey we proceeded to establish a
representative sample size – and to distribute the sample
across our 38 administrative units. Our selected Thana
administrative units comprised 115,000 households, or
just over half-a-million members. We applied standard
sampling and Probability Proportional to Sample Size
and segmentation procedure. This guaranteed that each
individual (child aged 6–14) in the population of children
in the 38 domains selected had the same probability
of being sampled. For each unit, the population aged
6–14 was clustered into segments of 100 children. These
clusters formed our Primary Sampling Units. We listed 150
households in 30 clusters, creating a total of 4,500 listed
households for which we collected data on household size,
composition, gender, education, the number and gender
of children aged 6–14 and whether they were engaged in
any job. We computed the sample size for the number of
children to be interviewed on a randomised basis at 2,700
(Annex 1).
Step 4: Survey implementation: We developed a
household listing from which the sample of 2,700 children
identified in Stage 3 could be selected. These children
are distributed across 23 parts and sub-parts in the 38
administrative units (Table 3).
40
30
10
Thanas
Child labour and education – a survey of slum settlements in Dhaka  33  
Figure 11. Adult illiteracy in Bangladesh’s Thanas

50
40
% of illiterate persons

40
30
30
20
20

10

Thanas

Note for figures 10 and 11: Thanas in red are those included in the survey.
Source of data for figures 10 and 11: Bangladesh Bureau of Statistics, 2011 Population and Housing Census.

Table 3. Sample selection: 23 parts and sub-parts in 38 administrative units

Total No. of segments

Thana Ward/Union Part Sub-part
for each domain
Adabor Thana Ward No-43 Total Comfort Housing Society 1
Ward No-43 Total Nabinagar Housing 1
Ward No-43 Total Shyamoly Housing Society 1
Badda Thana Badda Union Total Badda Dakshin Ananda Nagar 1
Bhatara Union Total Bhatara Nayanagar 1
Satarkul Union Total Sutibhola (Part-1) Padaridia Purba 1
Kafrul Thana Ward No-04 Total Baishtek 1
Ward No-15 (Part) Total Dhamalkot 1
Ward No-15 (Part) Total Lalashari 2
Ward No-16 Total Dakshin Ibrahimpur 2
Kamrangir Char Thana Sultanganj Union Total Hashlai 1
Khilgaon Thana Ward No-26 (Part) Total Meradia(Part-1) 3
Ward No-26 (Part) Total Meradia(Part-2) 2
Dakshingaon (Part) Union Total Nandi Para(Part) Paschim Nandi Para 1
Lalbagh Thana Ward No-91 Total Hazrat Nagae 1
Ward No-92 Total Islamnagar 1
Ward No-92 Total East Baragram 1
Ward No-92 Total Mominbagh 2
Shah Ali Thana Ward No-08 Total Mirpur Sec-1 (Block-H) 1
Ward No-08 Total Uttar Bishil-Kha 1
Tejgaon Ind. Area Thana Ward No-37 Total Begunbari 2
Ward No-37 Total D.Paschim Tejgaon Ind.Elaka 1
Ward No-37 Total Kuni Para 1

34  ODI Report

 The selected children constituted the respondents to
a detailed survey questionnaire developed with technical
advice from Understanding Children’s Work and BRAC
Institute of Educational Development, BRAC University
(BIED, BRACU). At the time our survey was conducted
there was no updated household census for slums in the
DCC area. The 2014 Census of Slums was not specified to
create representative population samples for the 6–14 age
group. In the absence of detailed census information, we
created our own household listing for 4,500 households
with 18,522 household members. The survey was
conducted over a six-week period between November
2015 and January 2016. We recruited and trained 35 Field
Research Assistants (FRAs) or enumerators to carry out
the survey. The FRAs operated in teams of two or three
members, spending on average three days in a segment
for household listing and 15–20 days for interviews.
Enumerator training, robustness checks and survey
supervision details are provided in Annex 1.
The limitations of the survey methodology have to
be acknowledged. Our survey was restricted to children
living in the defined enumeration areas. For that reason,
it does not capture work carried out by children who are
living-in as domestic workers. Given that the vast majority
of children working in this sector are young girls, it is
probable that we underestimate both the incidence of
child work in general and the incidence of female child
work in particular. Other aspects of our survey design
may also obscure important gender effects. Because we
focus on employment outside the home, we do not capture
work carried out by children within the household on
domestic chores, sibling care and other unpaid activity.
This is another area in which the work burden falls
disproportionately on young girls (UCW, 2011). Finally,
several studies have documented large numbers of children
living on the streets of Dhaka (INCIDIN Bangladesh,
2006; FREPD, 2003). While the numbers are uncertain,
many – perhaps a majority – of these children will be
working, often in hazardous activities. Once again, our
residency requirement means these children are not
covered.
Other limitations are inherent in the type of exercise
undertaken through the survey. Establishing the age
of children was difficult because only a minority of
children in the slums we covered (around 7.8%) have
formal birth registration. We addressed this problem by
training enumerators to work with children and their
parents in establishing age through critical timelines on
‘events calendars’. In our survey we also seek to capture
the perceptions and experiences of children themselves.
Inevitably, perception and subjective experience is difficult
to calibrate against a common benchmark. For example,
children may experience the same circumstances – such
as risk at work or the attitude of a teacher at school – in
very different ways. In developing our survey, we made
a point of framing questions around the categories used
by children themselves and reporting on their responses
through the same categories.

Box 2. Dhaka City Corporation – ‘North versus South’

In 2011, the Government of Bangladesh amended the Local Government City Act to divide Dhaka City
Corporation into Dhaka South and Dhaka North. While the slums selected for our survey have many features in
common with respect to population density, housing quality and social deprivation, there is a partial North–South
divide.
Dhaka South broadly corresponds to the administrative centre of Old Dhaka. Informal settlements are often
better established than in Dhaka North. In part, this is because of the pre-existing population density and size of
the slums. The CUS 2006 survey identified Kamrangir Char, located on the Buriganga River south of the University
of Dhaka, as one of Dhaka’s largest slums. The settlement grew rapidly during the 1990s with the expansion of the
ready-made garment sector and a vibrant informal economy. Immediately adjacent to Kamrangir Char, Lalbagh
Thana includes a dynamic small-scale informal sector working in metals, embroidery products and bicycle parts,
along with an established ready-made garment sector.
Dhaka North is in many respects the slum growth story of the past 20 years. The main commercial areas such
as Mahakhali, Gulshan, Kawrbazar and Uttara have grown rapidly, driven by the expansion of the ready-made
garment industry and garment-buying houses, a significant hotel and restaurant trade, banking and informal
sector workshops. The Tejgaon Industrial Area, one of our survey sites, houses large-scale formal sector industries
in areas ranging from garments to pharmaceuticals, foodstuffs, ceramics and chemicals, and has attracted a large
migrant workforce. Expansion has in turn created an expanding informal economy and retail sector. Kafrul Thana
houses a large informal handicraft sector, including embroidery, along with established ready-made garment
producers.

Child labour and education – a survey of slum settlements in Dhaka  35  

3.1.1 The social panorama of slums – differences
matter
The term ‘slums’ conjures up an image of uniform
deprivation. That image can be misleading. Slum dwellers
are uniformly deprived relative to some social groups.
However, in Bangladesh as in other countries, there are
marked differences between and within slums (Cameron,
2010; 2011). These differences are strongly associated with
education opportunities and with the risk of premature
entry into the world of work.
The slum areas covered in our surveys reflect the wider
national picture. With the exception of Tejgaon, slums in
Dhaka South have more settled habitation patterns than
those in Dhaka North, reflecting the rapid growth of the
latter (Box 2). The average period of residency reported
by households ranged from six years in Badda to 14 years
in Khilgaon (Figure 12). Patterns of migration also vary.
The majority of households included in our listing moved
to their current location from a prior settlement in Dhaka.
Around one third came from urban areas outside Dhaka
(Figure 13). Once again there were marked differences in
patterns of migration across slum areas (see Annexes for
details on specific slums).

Figure 12. Slum residency varies: average number of years the child’s household has lived in the slum

16 13.9
13.7
14
11.8
12 10.1
Average number of years

10
7.8
8 6.3 6.4
6
4
2
0
Badda Kafrul Adabor Shah Ali Tejgaon Kamrangir Khilgaon
Industrial Char and
Area Lalbagh
Dhaka North City Corporation Dhaka South City
Corporation
Note: Thanas in red are those included in the survey.

Figure 13. Slums have different patterns of migration: location from which the household moved prior to residency

80
69.1 69.3
70 61.8
59.6
57.0
60 54.2
50.3
50 43.7 45.0
39.5
% 40 32.8
27.4 25.2 26.9
30
20
9.5
10 2.1 2.7 0.8
3.5 1.2 0.8 2.7 5.0
2.6 3.6 2.4 1.4
0.0
0
Adabor Badda Kafrul Shah Ali Tejgaon Khilgaon Kamrangir
Industrial Char and
Area Lalbagh
Dhaka North City Coroporation Dhaka South City
Corporation

Other place in Dhaka Rural area (out of Dhaka) Urban area (out of Dhaka) Same slum

36  ODI Report

 The areas covered in our slum survey – The Haslai slum
in Kamrangir Char is dominated by housing structures
comprising tin and bamboo shelters. Because Kamrangir
Char is a port area for goods entering Dhaka South, slum
residents are actively engaged in loading and unloading,
working as porters (mutegiri), street vending for fruit,
vegetables and cooked food, with some also working in
garment factories. In the case of Kilgaon, there are no
garment factories located in the immediate vicinity of the
six areas covered. Livelihoods are dominated by activities
such as rickshaw pulling, hawking, bus assistant, and
domestic work.
Garments figure more prominently in the livelihood
patterns of slums in Dhaka North. Our slum survey sites
in Kafrul and Shah Ali are adjacent to Mirpur, which
is a centre for small-, medium- and large-scale garment
factories. Most of the residents in the three slum sites
covered in Tejgaon Industrial Area also report working in
the garment factories around the slum sites. Residents in
one of these slums – Dakshin-Paschim – have constructed
homes by hanging polythene sheets from bamboo. Access
to clean water is limited and sanitation non-existent. In the
neighbouring Kunipara slum, most shelters are constructed
from corrugated tin.
Residents in all of the survey areas face hazards
common to slum residents across Bangladesh. The Badda
slum areas have a major drainage channel running through
them and one – Dakshin Ananda – has a mini-lake within
We created a wealth index to explore the differences
and overlaps. Using a Principal Components Analysis,
we ranked households by access to sanitation, housing
structure, population density and asset ownership. We
then compared the distribution of asset-based wealth to
income distribution. Unsurprisingly, inequality levels in
the slums are less marked than they are for the country as
a whole. In part, this is because of how children’s wages
can reduce income inequality. However, as illustrated
in Figure 14, wealth is more unequally distributed than
income in the slums covered in our survey. This is reflected
in a higher Gini coefficient for wealth (0.30) across
our sample than for income (0.27) (Annex 2). Wealth
disparities may be significant in terms of participation in
education. Households with more assets may be better able
to withstand exogenous shocks – for example, a health
episode or losses of earnings – by drawing down resources,
rather than withdrawing children from school.
Figure 14. Wealth and income disparities in Dhaka’s slums
15
10
Household wealth index

the slum. Even light rain causes flooding. Slums in Adabor

are located close to, but on the wrong side of, a major
embankment protection system. The two slum sites in
5
Kafrul are located close to a lake and subject to constant
water-logging.
Household wealth and income levels vary across our
households, as does the interaction between the two.
0
All slum dwellers are deprived on some indicators of
0 50,000 100,000 150,000
well-being, but some are more deprived than others.
Comparisons between household income and wealth are
instructive. Income is an important indicator of deprivation
-5
– and it was striking that children in our survey define their Total household income
disadvantage principally in terms of monetary poverty. (Bangladeshi takas)
However, slum dwellers also define wealth and well-being
in broader terms to encompass shelter, access to services,
assets and security.

Child labour and education – a survey of slum settlements in Dhaka  37  

3.2 Child labour – incidence, profile and out-of-school and not in work. For children aged 11–14,
characteristics or junior secondary school age, 30% are reported as only
Are children living in slum areas working, in school, working, with boys entering labour markets in larger
combining school with work, or out-of-school and not numbers. One third of boys and just over a quarter of girls
in employment? Our survey addresses these questions by in our sample are reported as only working (Table 4).
drawing on respondent data from the 4,500 household
listing. As part of the listing exercise, we ask parents to Table 4. Education and employment status by age cohort:
report on children’s current employment and education children aged 6–10 and 11–14
status. This section provides a summary of the findings (the
Annexes provide additional data, including a breakdown
by location). Children aged 6–10 Children aged 11–14
(%) (%)
Before presenting the results, a word on terminology is
in order. Our definition of ‘work’ covers paid employment Boys Girls Boys Girls
outside the home and employment in a family business, Only studying 75.8 80.9 52.8 56.6
whether paid or unpaid. It does not include household
Only working 3.9 1.7 33.7 26.8
chores. This leads to an understatement of overall work
levels, especially for young girls. Studying and working 0.4 0.3 2.6 1.6
The results of the survey are striking (Figure 15). Across Doing nothing 19.8 17.1 10.8 15.0
the 6–14 age group, just over two thirds are attending
school. Another 16% are neither working nor in school. This does not imply that primary school age children
Working children account for 15% of the survey responses. are not working. Breaking down age cohorts into discrete
Almost all these children are ‘only working’ rather than years points to a marked increase in children working
combining work with school. Boys are more likely than during the upper-primary school years. Our household
girls to be working, and correspondingly less likely to be data points to 9% of 10-year-olds and 15% of 11-year-
in school. The fact that just under one third of the primary olds who are only working. By the age of 14 almost half
and lower secondary age cohort in a large sample survey of the children covered in our survey were in this category
Dhaka’s slums is either working or out-of-school is clearly (Figure 16).
a cause for concern. The fact that the incidence of working children rises
Disaggregating the age data provides insights into the from the age of 10 should not be interpreted as evidence
relationship between work and school. Consistent with that slum children complete primary school. Education
the picture presented by wider national surveys, there is in slums is marked by late entry. One third of six-year-
a marked age gradient for children in work. Relatively olds and a quarter of seven-year-olds in our sample were
few children enter paid work outside the home before reported as being out-of-school, implying that many
the age of nine. Correspondingly, most children in the start school at the age of eight or nine. Repetition is also
primary school age group are either attending school or common. The upshot is that many children are not in

Figure 15. School and work in Dhaka’s slums: household listing data on education and work status for children aged
6–14

80 71.0
66.4 68.6
70
60
50
% 40
30
20 16.1 14.1 16.2 16.2 16.2
11.9
10 1.3 1.1 0.8
0
Only studying (N=2,712) Only working (N=558) Studying and working Neither working nor
(N=43) studying (N=640)

Boys Total Girls

38  ODI Report

the right age-for-grade and that completion of primary
education by the age of 10 is the exception rather than the
rule. We return to these issues in Section 3.4.
Gender disaggregated age data illustrates the marked
differences between boys and girls in labour market entry
patterns (Figures 17a and 17b). An average 10-year-old
boy is twice as likely to be only working, with a marked
increase in the incidence of child work occurring between
the ages of 10 and 11. For girls there is a big increase in
the transition to work between the ages of 11 and 12, with
another surge between the ages of 12 and 13. By the age of
14, working rates for boys are still seven percentage points
above those for girls.
These distinctive patterns help to identify some of the
critical points for public policy intervention. Between the
ages of 10 and 12, around one fifth of girls and one quarter
of boys in the slum make the transition to the world of
work. Keeping all children in school for a full primary
cycle would dramatically change this picture, especially
given the widespread practice of late entry to school.
One of the most marked findings to emerge from our
survey is the segmentation of school and child work. As
noted in Section 2, previous studies based on national
data have found a significant number of working children
combining school with employment. For example, the
2013 Child Labour Survey found that 25% of child

Figure 16. Age gradients for school and work status: children aged 6–14

90 84.7
83.4
79.3
76.9
80
72.0
67.4
70
60.7
60
50.4
50 45.2
39.0
% 40 34.7
32.6

30 24.0 22.7

20 14.5 14.1 12.9 13.0

11.9 12.6 12.7
8.5 11.2
10 4.3
2.5 2.3 2.7
1.2 0.4 1.0 0.9
0
Only studying Only working Studying and working Neither studying nor
working

6 7 8 9 10 11 12 13 14

Box 3. Out of school and out of work

Children neither working nor in school are sometimes defined for survey purposes as ‘idle’. The term is misleading.
In-depth interviews with non-working out of school children in our survey sample – 16% of the 6–14-year-olds
covered – found that ‘idleness’ was the exception to a rule of children effectively forced out of education and
engaged in household chores. The cases of two children are illustrative (children’s names have been changed to
protect children’s identity).
Emon is a 14-year-old boy living in Mohammadpur Thana. He went to school when he was younger but has
learning difficulties. These are associated, in his father’s account, with autism. There are no schools in the slum
equipped to meet Emon’s needs – and his family is unable to afford the fees that would be required to send him to
a school equipped to provide specialised support. Emon works with his father during the day.
Sharmin is a 12-year-old girl living in Adabor Thana with her parents and two sisters. She was born in the slum.
After completing Grade 5 with a top score in the Primary Education Completion Exam, she had to stop studying
because her family faced financial problems. While the family’s position has now improved and Sharmin got
admitted into Grade 6, she cannot go back to school. With her mother and elder sister working she has to take
care of her younger sister and of her father, who is ill and unable to work. She now spends the entire day doing
household chores. Sharmin recognises the consequences of not attending school. ‘Nothing is possible without
education,’ she says, adding: ‘Education is needed everywhere, wherever you go.’ Her prospects are not promising.
With her elder sister about to marry, the family will lose an important source of income and Sharmin will have to
work. She is now trying to find work in the informal embroidery or formal garment sectors.

Child labour and education – a survey of slum settlements in Dhaka  39  

labourers in urban areas were attending school.18 By
contrast we find reported levels for children combining
education and employment at just 1% for children aged
9–10 and less than 3% to age 14. For slum children and
their parents there is a stark choice between school and
working, largely as a consequence of economic pressure.
Children reported as being out-of-school and out-of-
work are sometimes described in survey reporting as ‘idle’
(see Box 3). The depiction is unhelpful. Many six- and
seven-year-old children reported in this category are late
entrants to primary school. This is reflected also in the
peaking of the ‘only studying’ category at the age of eight.
Given that this is effectively the school starting age for
a significant group of children, an obvious implication
is that many reported as working by the ages of 10 and
11 have received just two to three years of education.
Older children – especially girls – are the ones likely to be

Figure 17a. Working, schooling and out-of-school without work: reported status of boys (4,500 household listings)

100
11.9 10.5 7.7
15.3 13.7 12.4 12.4
90 23.5 1.1 3.8
2.2
0.4 1.7 2.7
34.8 1.0 6.0
80 0.4
11.5
20.3
0.4 28.2 35.2
70
60 48.2
%
50
40 83.7 79.8
75.8 75.5
65.3 65.5
30 57.4 55.0
20 36.7
10
0
6 7 8 9 10 11 12 13 14

Only study ing Only working Study ing and working Neither working nor study ing

Figure 17b. Working, schooling and out-of-school without work: reported status of girls (4,500 household listings)

100
9.9 10.4
13.0 12.8 15.6 13.7
90 21.8 0.5 0.9 17.8
1.4 2.3 5.0
30.3 8.2 0.9 2.8
80 2.4
19.0
70
34.1 41.9
60
% 50
85.7 87.3 83.8
40 78.2 79.0
69.7
64.5
30
45.7
20 41.5

10
0
6 7 8 9 10 11 12 13 14

Only study ing Only working Study ing and working Neither working nor study ing

18 See Table 5.2.1, page 76 of the 2013 Child Labour Survey. The UCW country report for Bangladesh estimated lower numbers for children combining
school and employment, with 5% of children in employment in Dhaka combining both activities (and a national average of 6.8%).

40  ODI Report

performing household chores, or to have been pushed out
of school by circumstances beyond their control (Box 3).
Behind the average survey picture there are some
marked differences across the seven slum areas covered
in the survey (Figure 18). The incidence of working
children ranges from 9% in Khilgaon to 19% in Kafrul.
Similarly, the share of children reported as out-of-school
and non-working ranges from 8% in Shah Ali to 30%
in Adabor (where almost half of children aged 6–14 are
out-of-school). Our survey data and associated analysis
does not look at the distinctive drivers of children’s work
and education in specific slums. However, the results
underscore the potentially important roles of labour
market conditions and the social characteristics of
households in drawing children out of the world of school
and into the world of work.
3.2.1 The time intensity of work
In our survey of 2,700 children (as distinct from the
household listing) we asked respondents a series of
questions about their hours of work. What emerges is
a pattern of work in which the median child worker is
working at levels well above the 42-hour threshold set by
national legislation for ‘hazardous work’.
On average, children aged 11–14 report working
10.3 hours a day for six days a week. Only 15% report
working fewer than 42 hours a week – the threshold under
national legislation for hazardous labour. The median child
worker in our survey reports 70 hours in employment – an
extraordinary burden. Figure 19 captures the distribution
of daily hours of work. It shows 40% of children working
12-hour days.
3.2.2 Child work versus child labour
One important question is the degree to which the patterns
of child work revealed in our survey constitute ‘child
labour’. This question is critical for education policy,
employment law and child welfare policies in Bangladesh.
It also has a bearing on Bangladesh’s compliance with
wider international child rights provisions, including the
ILO Conventions on child labour and the Convention on
the Rights of the Child.
Although our survey does not allow for a detailed
assessment, the overwhelming bulk of the child work we
document would appear to fit squarely into both national
and international definitions of child labour targeted
for progressive elimination. For the slum children in our
sample there is effectively no border between child work
and child labour.
Consider first the national legislative context. For
children aged 6–11, working for any period of time in
employment outside the home is defined as child labour.
Working rather than attending school is also inconsistent
with legislation on education. Primary education is
compulsory between Grades 1–5 for any child aged
6–11. However, there is also a grey area with respect to
compulsory education. Given that many children start
school late and, even with smooth progression across
grades, would not complete school before they were
between 12 and 14 years old, there is a question over
whether compulsory education law should be applied by
reference to age or grade completion. Whatever the precise
interpretation, fully one fifth of 10-year-olds and one
quarter of 11-year-olds are either working or not attending
school.
Employment legislation sets some clear parameters. As
noted in Section 2, Bangladeshi law makes a distinction

Figure 18. Education and work status by slum location, children aged 6–14

100
100%
9.2 7.8 12.1 12.9
90
90% 21.1 20.4
9.4 14.7 29.5
80
80% 16.4 16.5
70
70% 13.1 19.1
60
60% 17.0

%50%
50
40
40% 81.4 77.5
71.5 70.6
30 65.8
30% 60.5
53.5
20
20%
10
10%
0
0%
Khilgaon Shah Ali
Shah  Ali Badda Kamrangir Char
Kamrangir  Char   Tejgaon
Tejgaon   Kaf rul Adabor
and Lalbagh Industrial  
and  Lalbagh Industrial Area
Area

Only
Only  studying
s tudying Working (including studying
Working  (including  s tudying  and
and  working)
working) Neither working nor
Neither  working   studying
nor  studying

Child labour and education – a survey of slum settlements in Dhaka  41  

for children aged 12–17 between involvement in ‘light
work’ and non-hazardous employment for up to 42 hours
a week, and child labour defined as working for more than
42 hours a week. Even without considering the distinction
between hazardous and non-hazardous employment, most
of the children in our survey constitute child labourers.
Only 15% report working fewer than 42 hours a week.
This suggests that the provisions of the Bangladesh Labour
Act of 2006 and the 2013 Children’s Act hold limited sway
in Dhaka’s slums.
The situation of slum-dwelling children is similarly
inconsistent with the principles underpinning international
child rights, as defined by ILO Conventions and the
Convention on the Rights of the Child. However,
inconsistency is not the same as illegality in this context.
The practices identified in our survey are not compatible
with ILO Convention 138, which allows ‘light work’
subject to the stricture that the education of the child is
not compromised. An overwhelming majority of working
children under the age of 14 in our survey are sacrificing

Figure 19. Working hours: reported weekly work hours, children aged 6–14

Average number of hours Median number of

100 worked in a week: hours worked
64 hours/child in a week:
90 70 hours/child
Cumulative percentage of children aged 6 –14

80

70

60

50

40

30

20 15.5

10

0
6
7

105
112
10
12
14
15
16
18
20
21
24
28
30
35
36
40
42
48
49
50
54
56
60
63
64
66
70
72
77
78
80
84
90
91
98 Number of weekly hours of work by children aged 6–14

Box 4. Working lives

Our survey provides data on a large group of children living in some of Dhaka’s slums. Behind the data are
individual stories, some of which we heard during interviews and focus group discussions in Adabar Thana, north
Dhaka (children’s names have been changed to protect children’s identity).
One of those stories belongs to Amina, aged 14. She moved to Adabar with her family seven years ago. Amina
completed Grade 4 of primary school but did not sit the Grade 5 exam. Her father fell ill three years ago and she
had to start working to help pay the medical bills. Amina now reports working 12 hours a day (with two short
breaks) as a domestic help, earning Tk 2,500 a month ($32). ‘I have lost a lot by not going to school. But my
family are poor and my father is sick.’
Shilpa, also aged 14, completed two years of schooling in a rural area before coming to Dhaka. She completed
Grade 5 in Adabar but now works in a garment factory operating a sewing machine. When asked about her hopes
for the future she comments: ‘I’m happy to help my family, but I don’t have dreams – and I will not be returning
to school’. She had a very positive experience of education, did well at school, and wishes she could resume her
education. ‘If I studied I could learn things, find a better job and earn more’, she says.
In a focus group discussion held with 15 working children aged 11–14 in Adabar, several recurrent themes
emerged. Almost all of the children reported having to leave school because of poverty and the need to earn
money. Most wished they had been able to continue their education. However, several reported leaving primary
school because of negative experiences, including abuse and beatings by teachers. As one 12-year-old boy put it:
‘I didn’t like school. Teachers did not respect me and I could not perform well. I was beaten and scared of going
to school’.

42  ODI Report

their education. Most are working well beyond the
threshold for light work recommended by the Convention.
However, Bangladesh has not ratified Convention 138.
3.2.3 Why and where are children working?
In this section we focus principally on the results of the
survey related to the world of work. Through the survey
questionnaire, we ask children about their employment
background and their experience of work. An important
caveat is that we did not conduct workplace surveys with
employers to corroborate the data, in part because of the
complexity of securing accurate reporting on what may be
illegal employment activities.
Most children report entering the world of work
before their adolescent years. The average age at which
working children in our survey start work is 11.3, with
girls typically starting a year later than boys (Figure 20).
Given that a large proportion of working children started
school after the stipulated entry age, and that many
repeated grades or dropped out at some stage, a significant
proportion of children start work before completing a full
primary cycle.
An overwhelming majority of both boys and girls
reported working to supplement family income, with
inability to afford school costs also figuring prominently
(Figures 21a and 21b). Almost three quarters of girls
Children were asked during the survey whether they
or their parents took the decision that they should start
working. Strikingly, a majority of both boys and girls
reported that they took the decision (Figure 22). Once
again, the data has to be treated with some caution.
Children and parents are clearly operating within highly
constrained parameters of choice and the lines in intra-
household decision-making are blurred. Parents who
send children to work because they are unable to feed
a household are clearly acting under compulsion (Box
5) – and children who concur in that choice in pro-actively
seeking work are not in any meaningful sense ‘free
agents’. Even so, many children see themselves as active
participants in the decision to start working
Figure 20. Working lives start early: reported age of entry
into employment for children who are working
14
11.8
12 11.3
10.9
10
Age in years

8
cited support for family income as the biggest factor (and
another 5% highlight help in paying family debt). Only a
small minority cite a lack of interest in school as a factor. 6
By contrast, just over half of boys cite supplementing
family income as the primary reason for their working – 4
but fully one fifth express a lack of interest in school as
the second most important consideration (Box 4). Boys are
2
also more likely than girls to feel that they may acquire
new skills through work.
0
Boys Girls Total

Figure 21a. Reasons for working: top five reasons cited by boys

Supplement family income 53.5

Not interested in school 21.3

Learn skills 7.5

Cannot afford school fees 4.9

Help in household enterprise 4.4

Help pay family debt 2.1

I am not a good student 2.1

0 10 20 30 40 50 60
%

Child labour and education – a survey of slum settlements in Dhaka  43  

Figure 21b. Reasons for working: top five reasons cited by girls

Supplement family income 73.6

Cannot afford school fees 8.8

Help pay family debt 5.3

Not interested in school 3.8

Learn skills 3.3

I am not a good student 1.5

0 10 20 30 40 50 60 70 80
%

Figure 22. Making choices: children reporting on who took the decision for them to start work?

57.0 55.1
60

50 43.3
42.0
40

% 30

20

10
0.3 1.3
0
By myself My parents/guardians Myself but I was not well-informed
and carefully informed on the
type of job

Boys Girls

Figure 23a. Cash payments dominate… Figure 23b. …And most children have no contract

If you are working for others' business, how is If you are working for others' business, have
your salary paid for? you been employed on the basis of...?
98.1
96.5 100 91.7
100 88.8 90
90
80
80
70
70
60 60
% 50 % 50
40 40
30 30
20 20
6.7 3.5 8.3
10 10 1.9
0 0
Cash Cash and in kind A verbal agreement A written contract

Boys Girls Boys Girls

44  ODI Report

 Our survey provides insights into the terms on which
children enter the world of work. Only a small group
of (predominantly male) children report working in a
family business. The vast majority report working outside
the household for cash and with no contract (Figures
23a and 23b). While this result is predictable in light
of the questionable legality and, in some cases, outright
illegality of the employment practices involved, it places
children in a highly vulnerable position with respect to
their employers. The fact that cash-based payment is the
primary source of remuneration for children may serve to
further limit their legal entitlements and the corresponding
obligations on the part of their employers.
3.2.4 Source of employment
Where are children working? We asked children to report
on their principal source of employment, either in terms
of employer or activity. Our questionnaire does not make
a generalised distinction between formal and informal
work. However, in the case of garments, respondents were
given a choice between ‘sewing clothes’ (informal sector)
versus ‘garment worker’ (formal sector). In cases where
respondents did not fully understand the distinction,
enumerators explained that formal sector garment work
involves employment by a large- or medium-sized factory.
Our survey does not consider whether informal garment
work involves subcontracting from formal employers or
not. Reported employment is highly segmented. Around
two thirds of girls report working in garment factories,
with another 10% working in sewing or handicrafts. For
girls living in slums, the ready-made garment sector was
by far the largest reported source of employment. Boys
report a more diverse pattern of employment. Just 13%
were working in the formal garment sector, and other
activities – street vending, work in shops, employment by
workshops, day labour – accounted for around half of
reported employment

Figure 24a. Where are children working: source of employment reported by boys

Grocery shop 14.8

Garment worker (formal sector) 13.3
Salesman 12.7
Work at workshop 9.3
Day labourer 8.6
Sewing clothes (informal sector) 5.6
Crumb collector/seller 4.7
Vegetable seller 3.4
Work at hotel 3.3
Driver helper 2.9
Handicraft maker 2.2
Shoemaker/polish, repair shoes 2.2
Construction worker 2.1
0 2 4 6 8 10 12 14 16
%

Figure 24b. Where are children working: source of employment reported by girls

Garment worker (formal sector) 62.4

Saleswoman 11.2

Handicraft maker 5.0

Domestic worker 4.9

Sewing clothes (informal sector) 4.6

Day labourer 4.0

0 10 20 30 40 50 60 70
%

Child labour and education – a survey of slum settlements in Dhaka  45  

Figure 25.CWES
Figure 25. CWESslum
slum locations
locations andand garment
garment and apparel
and apparel factories
factories in Dhaka
in Dhaka

Badda Thana
Shah Ali Kafrul (part)
Thana Thana

Badda Thana
(part)

Tejgaon Ind.
Area Thana
Adabor
Thana Khilgaon
Thana

Lalbagh Thana

Kamrangir
Char Thana

Registered with the Alliance or the Accord

Registered with the Government or the trade associations

Bangladesh

11,824 workers
Dhaka

200 or fewer workers

 The prominent role of garment factories is one of
Figure 26. The formal garment sector is reported as a
major employer: age profile of 10–14 year-old children
the more surprising findings to emerge. It is widely
reporting work in the formal garment sector
assumed that the sector is more intensely regulated
than other sectors. It might have been expected that the
prospect of investigation and monitoring would deter 100
the recruitment of under-age children. Child labour 90
legislation is enforceable through fines – and the Chief 34.5
Inspector of Factories (which operates under the auspices 80
48.3
52.9
of the Ministry of Labour) is empowered to carry out 70
unannounced spot checks. 60
Our survey suggests there may be a systemic problem %
of enforcement and compliance (Box 6 and related Figure 50 34.5

25). Figure 26 provides an age profile for employment 40

32.8
in the garment sector, as reported by children. Over one
30 32.2
quarter of the boys employed in the sector are aged 12 or
under. Consistent with the wider gender pattern, girls on 20 22.2

average report entering garment sector employment later. 10 10.1

13.1
2.4
Just under half of those reporting working in the garment 6.4 2.6 2.6
0 2.2 3.2
sector were aged 13 or under. Once again, many of these
Boys (40 Girls (148 All children
children are working longer than the 42 hours per week
observ ations) observ ations) (188
threshold stipulated in legislation. observ ations)
Various factors contribute to the regulatory challenges.
The fact that children do not have contracts and are paid 10 11 12 13 14
in cash makes investigation difficult. During in-depth
interviews, employers themselves drew attention to the There are also marked differences in the profile of child
difficulties associated with establishing the age of job work across our survey locations (the Annexes). In the
applicants given the large numbers of young people lacking slum areas of Dhaka North, the formal garment sector is
birth registration (Box 7). Focus groups involving children overwhelmingly the largest employer, accounting for over
reported a generalised failure on the part of employers to half of total employment in Kafrul and Shah Ali. Street
request age information. vending figures far more prominently in Kamrangir Char,
while the Tejgaon Industrial Area combines a range of
formal and informal sector activity.

Box 5. The parents’ perspective

Two accounts from parents in the Kafrul slum help to illustrate the economic pressures that drive children into the
world of work (parents’ names have been changed to protect their identity).
Mohammed Jamal moved to Kafrul in 2013 after he was unable to pay back a loan taken out following
flooding and the loss of a crop on his farm. He is currently a day labourer. After moving to Dhaka, Mr Jamal
reports that he had no choice but to send his children to work. He cites high rent and living costs and the cost of
education as barriers. One of his daughters, now aged 12, found a job through relatives in a garment factory, but
was fired after it was discovered she was under-age. She then found a similar job in another garment factory. Mr
Jamal reports that she works 12 to 13 hours a day and for six (sometimes seven) days a week. She earns
Tk 8,000 a month – and Mr Jamal sees this as the minimum he would need for her to return to school. He thinks
that working will be beneficial for her future, because it will make it easier to arrange a ‘good marriage’. Mr Jamal
recognises that dropping out of school comes with costs. He believes that, if his daughter secured a Secondary
School Certificate, she would get a better salary.
Mina Akter is a mother of six children, two of whom – one boy and one girl – are working. Ms Akter is
employed as a domestic worker in middle-income households. She is illiterate, never having attended school.
Her working son, aged 14, is a crumb collector and seller (known as Tokai and Vangari trader in Bangla). He
works eight/nine hours a day for five/six days a week. Ms Akter says she has no choice about allowing her son
to work because the costs are unaffordable. Asked about the implications for her son of not attending school,
Ms Akter says: ‘Nobody values an illiterate person’. She knows that, without education, he will have very limited
opportunities in his future.

Child labour and education – a survey of slum settlements in Dhaka  47  

 Box 6. Garment exports and child labour
Our survey raises serious concerns over the issue of child labour in the supply of garments from factories in
Bangladesh to consumers in Europe, the US and elsewhere. We did not carry out a detailed review of individual
factories, but the sheer scale of child employment in the sector – and the links between small-scale factories and
large-scale exporters – make it highly probable that children in Dhaka are involved in export production.
New York University’s Stern Center for Business and Human Rights recently undertook detailed mapping of
garment and apparel factories in Dhaka and other cities. Based on large-scale data analysis of identified factories
in Bangladesh in 2015, the Center estimated 7,000 factories, employing around 5 million workers, producing for
export.
Direct suppliers account for around half the sector’s factories. These receive licences to import apparel and
fabric duty-free for manufacture into export products. They are generally large-scale operations that have
contractual relationships with foreign brands and input suppliers, and access to export credits and significant
capital investment.
Indirect suppliers play a critical role in the export supply chain. These factories contract with direct suppliers,
enabling the latter to adjust to shifts in demand. Around half are formal sector operators that have registered with
trade associations like the Bangladesh Garment Manufacturers Association (BGMEA). Around 1,000 of these
factories produce for export through sub-contracting with direct exporters.
Informal factories are a sub-set of indirect suppliers. They do not register with government and trade
associations. They rely heavily on sub-contracts with larger factories to fill their production lines. Of the 479
factories surveyed by Stern School researchers in 2015, around one third were informal sub-contractors. On
average they employ around 55 workers, often focusing on a single specialised process, such as sewing, washing,
dying or printing.
Workers in informal factories are highly vulnerable. The sector operates on thin margins and the monitoring
of safety standards and labour rights is weak to non-existent. While the Stern Center survey did not report on
child labour, the survey teams did observe child labour in informal factories. While precise figures are impossible
to establish, it is likely that many, if not a majority, of child workers enter the sector through informal enterprises
before moving into the formal sector.
Informal factories are heavily engaged in export production. In 2015, 91% were producing wholly or partly for
export, with most sub-contracting for producers supplying national and international markets.
All our survey sites are close to both formal and informal factories, and one site, Mirpur, has one of the highest
concentrations of registered factories in Bangladesh. The survey results add to a wider body of evidence that points
to systematic employment of children in the garment sector.
We do not identify named foreign brands that can be linked to factories employing child labour. However,
it stretches credibility to assume that the supply chains for these brands do not include significant employment
of child labourers. First, with over 60% of working girls and 13% of working boys reporting employment in
the garment industry, this is the single largest source of employment for children in our sample. Second, the
operations of direct and indirect exporters, and formal and informal suppliers, are deeply integrated. Widespread
sub-contracting renders it highly probable that children are producing clothing destined for international supply
chains.
The vast majority of these children will be working in factories that provide limited protection. Two factory
programmes – the Accord and Alliance programmes – have been established by foreign brands. However,
they cover only around one quarter of Bangladesh’s garment factories. Most of these are large, formal sector
enterprises, constituting around 60% of factories on the direct export list. However, even this group is extensively
involved in sub-contracting, and there is limited reporting on their suppliers.
Even within the elite safety group there are serious concerns over the effectiveness of regulation. As of October
2015, only eight Accord/Alliance factories had passed final inspection. The vast majority of those inspected had
been required to adopt Corrective Action Plans, few of which have been successfully completed. Very few indirect
exporters have been subjected to safety inspection.
Foreign brands could do far more in terms of constructive solutions to the child labour problem. As well as
requiring direct suppliers to provide more and better information on their sub-contractors, they could actively
support efforts to comply with higher safety standards. While ultimate responsibility for strengthening the
regulatory regime rests with the Government of Bangladesh, brands could – and should – be creating incentives for
firms to comply with child labour laws.
Source: Labowitz and Baumann-Pauly, 2015.

48  ODI Report

3.2.5 Control, risk and hazard in the workplace
Assessing the conditions experienced by children at work
is inherently difficult. Our survey asked children to report
on the attitude of employers, along with a range of proxy
indicators for risk and hazard in the workplace. The
findings have to be interpreted with some caution. Children
working in different sectors may have very different
experiences and perceptions of risk which are difficult to
compare on a common scale. The questions framed in our
survey were based on the concerns identified by children
themselves in a pilot exercise – and we made no attempt to
evaluate the self-reported statements against a unified risk
and hazard benchmark. It is also worth emphasising that
children may have a limited awareness of their rights and
the protection afforded by legislation.
Despite these reservations, the survey results highlight
a number of concerns. Most children self-report that their
employers treat them well. However, around 13% of boys
and 20% of girls report being treated either badly or very
badly (Figure 27). Of the children reporting bad treatment,
shouting and otherwise insulting behaviour was identified
as the greatest concern. However, recourse to physical
beating was also cited as a feature of employers’ bad
behaviour (Table 5).
Table 5. Children reporting compulsion in the workplace,
selected practices
Practices identified by children: the
Boys (%) Girls (%)
employer sometimes compels us to…

Figure 27. How is the child treated at work: perceptions Work extra time 19.2 37.2
reported by working children Work without payment 4.5 6.9
Be available any time for work 6.8 6.3
90
Perform work that is not in the contract 6.3 13.2
80.2
80 77.1
Work for other employers 4.5 4.0

70 Being subject to freedom-less work 13.2 16.0

Percentage of children

Children reporting on consequences of failure Boys (%) Girls (%)

60
to comply with employer demands: I would be
50 subjected to
Subjected to physical punishment 12.5 12.4
40
Given more work 22.2 20.1
30
Denied food and rest breaks 3.9 0.9
20 15.7 Fired 29.7 33.1
10.8
10 6.7
2.3 4.0 3.2
0
Very badly Badly Well Very well

Boys Girls

Box 7. In-depth interview with the manager of a garment factory

Employers report difficulties ascertaining the age of children. One manager of a garment factory acknowledged
that he was employing ‘around 10’ children aged between 11 and 14. He explained that the recruitment process
involves posting vacancies on the factory notice board and, in the case of casual or temporary employment, a
‘factory gate’ interview followed by a medical examination. In principle, employees are asked to provide their
national identity card or birth registration card to verify they are over 18. In practice, though, the manager
indicated that many employees had neither card. While aware that children aged 11–14 should not be working, he
did not regard their employment as illegal.
Once hired, children receive an appointment letter, an attendance card and a factory photo identity. The
manager indicated that he did not allow children to work with heavy machinery. They were principally employed
in sewing clothes, cutting thread and checking material.

Child labour and education – a survey of slum settlements in Dhaka  49  

 Many children report demands that infringe their rights,
with a significant minority experiencing what amounts
to bullying by employers. Around one third of girls and
one fifth of boys report being forced by employers to
work extra hours. Girls are also more likely than boys
to be asked to perform work not agreed in their terms of
employment. We asked in the survey what would happen
were a child to refuse to carry out tasks assigned by the
employer. ‘Getting fired’ was the single biggest anticipated
response, followed by the prospect of being given
additional tasks and physical punishment.
Table 6. Percentage of children that experienced any of
the following…
Percentage of children that experienced
Boys (%) Girls (%)
any of the following…
Back pain 23.2 28.4
Extreme fatigue 36.1 34.6
Carrying heavy loads 78.8 21.2
Operating machinery/heavy equipment 44.2 55.8
Exposure to dust, fumes, toxic substances, 20.0 24.8
chemical substances
Exposure to loud noise or vibration 21.5 26.3
Exposure to extreme cold or heat 21.1 19.9
Use of dangerous tools (knives etc.) 8.2 4.0
Children report exposure to health risks and hazards
in the workplace. We asked respondents to report on their
experience of work-related injuries and ailments over the
preceding 12 months (Table 6). Significant minorities of
boys and girls reported extreme fatigue (consistent with
evidence on working hours), back pain and fever, along
with superficial injuries. Many children – especially boys
– appear to be expected to carry heavy loads. Moreover,
55% of girls and 44% of boys reported operating
machinery or heavy equipment. There are also high
levels of reported exposure to dust, fumes and chemical
substances, loud noises, dangerous tools and extreme
temperatures.
3.2.6 Levels of payment and expenditure priorities
Children overwhelmingly cite economic pressures as the
primary force drawing them into the world of work.
Combining the data on parental income with those on
children’s level of remuneration provides some valuable
insights into these pressures. One important finding that
emerged is that income from children’s work has the effect
of broadly equalising the average household income of
working and non-working children. This lends support to
the contention that parents allow children to work as a
distress choice aimed at meeting a subsistence threshold.
50  ODI Report
Table 7 summarises the average income position of
working and non-working children’s households. It
highlights the critical role of the father’s income as the
main source of divergence. Average reported parental
income in households with no working children was
Tk 13,000 compared to Tk 9,000 for households with
working children. The average income derived from the
work of children was Tk 3,600 ($47), bringing the income
of their households up to the average level of households
without working children. This lends prima facie evidence
to the argument that parents are forced by circumstance
to send children to work in pursuit of a minimum income
level. The reported monthly income of working children
is Tk 3,664, around $47. The role of child work in raising
the income of households to the average level for the
slum lends some weight to the view that recourse to child
labour reflects a concern on the part of parents to achieve a
minimum income threshold.
Table 7. Reported income by household status: average
income levels for mother, father and working children
Households of non- Households of
working children working children
Tk 11,582.63 Tk 6,634.14
Father’s income
($147.84) ($84.67)
Tk 2,332.02 Tk 2,438.46
Mother’s income
(£29.76) ($31.12)
Working children’s income NA Tk 3,663.82
(6–14-year-old child who is ($46.76)
the survey respondent)
Other family members’ Tk 3,201.31 Tk 4,999.06
income ($40.86) ($63.81)
Tk 17,115.96 Tk 17,735.48
Total monthly income
($218.46) ($226.36)
Currency exchange on 10 August 2016 (Tk 78.35=$1) from
http://www.xe.com/currencyconverter/
Our survey does not allow for a detailed review of
underlying labour market conditions. Perhaps reflecting the
dominance of formal sector garment employment, there is
a modest wage premium in favour of girls (Tk 4,177) over
boys (Tk 3,713) (Table 8). The gender premium raises an
obvious question: namely, if the returns from girls’ work
are higher, why do boys on average enter labour markets
earlier? The answer to this question could lie partly in
education or age entry requirements for the garment sector,
or in the perceived returns from females attending primary
school.
What is clear from our survey is the overwhelming
perception among children that their work is critical
to household welfare. When asked ‘how did you spend
your money last month’? around 90% of boys and girls
identified contributing to family income as the most
important item.
 The limitations of our survey with respect to reported
income have to be recognised. Children were asked to
report on their daily and monthly income, along with the
typical number of hours worked a day and number of
days’ work a week. In reality, it is likely that many working
children are paid at variable levels on an intermittent
basis. By reporting the observations of the child we are
almost certainly smoothing what could be an irregular
and erratic stream of income. This is especially true for
children involved in informal work and day labour. It may
be less true for children employed in the formal garment
sector. Inevitably, average figures also obscure variations
associated with age.
Table 8. Reported income by household status: average
income levels reported by child (Bangladeshi takas,
(US $))
While our data does not provide a comprehensive
picture of comparative labour market returns, it does
offer some insights into patterns of remuneration for
children working in the garment sector. There is an
established national minimum monthly wage for the
formal garment sector set at Tk 5,300. This is more than
the average reported wage (Tk 4,684) for the children in
our survey sample. However, the simple comparison may
underestimate the real gap. The national minimum wage is
stipulated for eight hours’ work a day for 26 days a month.
Expressed differently, they work 52 hours more for 13%
less income. Translated in equivalence against the national
minimum wage, they are working for an hourly rate of
Tk 14 an hour versus the Tk 25 required under minimum
wage legislation. In other words, they are receiving only
54% of the minimum wage (Figure 28).

Boys Girls
Daily Income Tk 130.69 Tk 143.68
($1.67) ($1.83)
Monthly Income Tk 3,731.76 Tk 4,177.64
($47.63) ($53.32)

Currency exchange on 10 August 2016 (Tk 78.35=$1) from

http://www.xe.com/currencyconverter

Figure 28. Wage of child workers in the formal garment sector in comparison with national minimum wage (in the
garment industry)

5,400 100
5,300 88.4
5,300 90
5,200 80
5,100 70
Bangladeshi takas

5,000 54.4
60
4,900
% 50
4,800
4,684 40
4,700
30
4,600
4,500 20

4,400 10
4,300 0
Average reported Minimum monthly Monthly wage of Hourly adjus ted wage
monthly wage of wage for the formal working children as of working children as
child workers in the garment s ector percentage of percentage of
formal garment national minimum national minimum
s ector wage in garments

Child labour and education – a survey of slum settlements in Dhaka  51  

3.3 Some children are more at risk of child Figure 30. Wealth matters: probability of working by age
and wealth index for a child living in slums
work than others
Slum-dwelling children are not a homogenous social
category. They come from households with different levels 30
of parental literacy, income and wealth. The composition 25.2
of their households vary. They live in settlements with 25
21.8
different levels of basic provision. All of this has a direct
bearing on the distribution of risk within slums of being 18.7

Probability of working
20
drawn into the world of work. Based on a literature review, 15.4 15.8
we identified a range of potential variables that might 13.2
15 12.9
explain the probability of a child working. We then ran a 10.6
probit model from which we derived marginal effects for 8.6
10 6.9
the probability of working. The following summarises the
main findings (see Annex 3 for technical detail).
5
1. There is a steep age gradient structured by gender. As
highlighted in the previous section, the probability of 0
Poorest Poorer Middle Richer Richest
a child working increases sharply from the age of eight
(Figure 29). By the age of 14, the children in our survey Wealth Index
have just under a one-in-two likelihood of working.
Measured in terms of absolute percentage difference, Girls Boys
the gender gap widens steadily to the age of 14. While
our survey does not address the underlying causes, there
would appear to be a set of labour market factors at
play interacting with parental perceptions of the relative Figure 31. More income is associated with less child
labour: probability of working based on monthly
value – and opportunity costs – of education for girls
household income (excluding income from the child
and boys.
labour)

Figure 29. The age gradient for working children:

30
probability of working by age and gender for a child living 26.3
in slums
25
18.9
70 20
Probability of working

16.4
60.0
60 15
10.9
50 46.1 10.6
Probability of working

10
40 41.1 5.4
32.4 5
30
20.7 27.9
0
20 1,000 15,000 120,000
11.9 17.2
6.1 Average monthly income for the child
10 2.8 9.5
0.4 1.2 household (Bangladeshi takas)
4.7
0 0.3 0.8 2.1
0.1
1 2 3 4 5 6 7 8 9 Girls Boys
Child Age

Total Girls Boys

52  ODI Report

2. Child work is inversely related to household income
and, more especially, wealth. There is a marked wealth
and income gradient within slums. Ranked on our
wealth index, male and female children from the poorest
quintile are twice as likely to be working as children
from the richest quintile (Figure 30). There is also an
income gradient for the probability of working, though
this is considerably less steep than the wealth gradient
(Figure 31). The risk of children working appears to
increase quite markedly at income levels below
Tk 15,000. Working children themselves are more likely
to self-identify as poor. In our survey, 76% described
themselves as poor compared to 56% of non-working
children. The gender gap converges slightly along the
gradient. This may constitute evidence that the parents,
defined in terms of their wealth, have the least capacity
to adjust to external shocks without taking children out
of school. If wages are higher and earning opportunities
greater for boys at lower age levels, they are likely to be
taken out of school first.
3. More recent settlers and migrants from rural areas
face disadvantages. Households that have been living
in a slum area for 10 years or more are less likely to
have children working rather than attending school
(Figure 32). Migrants from rural areas also appear to
face elevated risks of child labour (Figure 33). This may
reflect the economic consequences of distress migration
or inability to access education provision – or some
combination of both factors.
Figure 32. More settled residency patterns reduce child
labour: probability of working based on the number of
years child household has been living in the same slum
25 23.0
19.0
20
15.5
Probability  of  working
13.9
15 12.4
11.0
9.7
8.5
10 7.5
6.5
4.8
5 3.5
0
0 10 20 30 40 50
Number   of  y ears  t hat  t he  c hild  hous ehold  has
been  living  in  t he  s ame  s lum
Girls Boys
Figure 33. Arrival from rural areas brings elevated risks of
child work: probability of working based on rural/urban
relocation
25
21.6
20 18.0
Probability of working
15 12.9
10.4
10
5
0
Girls Boy s
From rural areas From urban areas
4. Household composition and characteristics matter.
Households with more adults and fewer child siblings
have more breadwinners, and are therefore better able
to afford the direct costs and the opportunity costs of
education. On average, working children come from
households with 5.1 children as compared to 4.7
children for non-working households. The probability
of a child working rises with the percentage of children
aged under 14 – and the effects are quite large (Figure
34). Living in a father-headed household also greatly
elevates the probability of a child working (Figure 35).
Households with parents reported as having a disability
and children who have been orphaned are marked
by a higher incidence of children working, reflecting
economic pressures. Around 10% of household heads
in our survey have a reported disability. However, 15%
of households with working children fall into this
category. Birth registration is another variable marked
by a strong association with employment status. Only
5.6 a small minority of children in our sample have a birth
4.1
3.0 registration card. However, working children were far
less likely to be registered (2.7% of our sample) than
1.7
1.2 non-working children (8.9%).
5. The education level of parents materially affects the
60 70 80
risk of a child working. Both maternal and paternal
 
education has an influence on the incidence of child
work. Having a mother or father with some secondary
education approximately halves the risk of a child
working relative to the child of a parent with no
education – the effects are marginally stronger for
fathers (Figure 36). A majority of the parents of
working children report never having attended school,
Child labour and education – a survey of slum settlements in Dhaka  53  
 so parental education is a powerful mechanism for associated with education and the advantages associated
transmitting the risk of child work. Only 8% of the with a home environment in which some level of literacy
mothers of these children report having completed prevails. We created a composite indicator based on
primary education, compared to 16% for mothers of parental reporting and perceptions of school, including
non-working children. Beyond serving as a proxy for cost, distance and quality. We then measured the effects
deprivation, parental education may influence whether of discrete improvements in individual components,
or not children participate in school through other which we rank by order of impact (Figure 3.29).
transmission channels. These range from the value Distance from school weighs heavily, reflecting perceived
placed on education, to perceptions of the benefits opportunity costs.

Figure 34. Households with fewer children are less likely Figure 35. Father-headed households are more likely to
to have children working: probability of a child working include working children: probability of working based on
based on the percentage of children younger than 14 whether father or mother household head (HH)

30 20 19.0
27.3
25.8
24.3 18
25 22.9
21.5
20.2 16
Probabilit y of working

18.9
20 17.6 17.3
16.4 14
Probabilit y of working
16.1
15.3 15.0
14.2 13.9
15 12.9 12 11.1
11.9
10.1
11.0 10.1
9.3 10
10 7.8 8.5
8
5
6 5.2

0 4
0 10 20 30 40 50 60 70 80 90 100
Percentage of children younger than 14 2
in the child's household
0
Girls Boys Girls Boys
Girls Boys Mother is the Father is the
household head household head

Figure 36. Parental education strongly influences the child work patterns: probability of working based on the
education level of the child’s mother and father

25
Probability of working

21.7
20.7
20 18.8 18.7
17.0
15.3 15.9
15 13.7 13.4
12.2 12.2 12.8
10.8 10.7 11.2
9.6 8.8 9.3
10 8.4
7.4 7.1
6.4 5.7
4.6
5

0
Girls Boys Girls Boys
Education level of the child's mother Education level of the child's father

Never attended school Some primary education

Completed primary education Some secondary education

Completed secondary education Completed graduation and/or master's degree

54  ODI Report

3.4 Schooling, learning and perceptions of
education
The impact of child work on education is one of the
most important transmission mechanisms for social
disadvantage. Children who are working rather than
building human capital through learning face a lifetime
of diminished opportunity. The links operate in both
directions. Children could be working because they are
out-of-school – and they could be out-of-school because
they are working. Whatever the precise direction of
causality in individual cases, children’s employment in
Bangladesh’s slums appears to have significant detrimental
consequences for their education.
Children in slums across Bangladesh suffer in acute
form from Bangladesh’s twin crises in education. They
are less likely to progress smoothly through a full cycle
of primary education into secondary school. Those that
do progress will struggle to achieve levels of learning
consistent with the ambition defined by the national
curriculum. Within slum areas, children who are working
or at risk of abandoning school for employment face a
double disadvantage.
Our survey casts into sharp relief the interplay between
household disadvantage and inequality in education
that effectively cuts short the school careers of so many
working children. It points to marked differences between
working and non-working children in terms of their ability
Figure 37. Probability of working for a child living in slums of Dhaka City Corporation based on existing education
opportunities
60
50.5
Probability of working
50
40
30 37.1
17.8
20
10
10.3
0 4.2
School far from School near home …with good quality
home, bad quality
teaching, high school
fees and high
schooling costs
Girls
to participate in school, progress across grades and learn.
The results suggest that child labour remains a major
obstacle to the realisation of the national goals set for
education in Bangladesh.
In this section, we present the survey data on the
experiences reported by children in school, children who
are working, and children who are neither working nor
in school. We do not report separately on the number of
children who are combining employment with education,
due to the very small group.
3.4.1 School attendance and grade progression
Attending school at the age corresponding to the standard
grade is the exception rather than the rule among slum
dwellers in our survey. Fewer than one in five children
aged 6–10 and just over 1 in 10 of the 11–14 age group
are at the expected age-for-grade. Figure 38 provides a
detailed age-for-grade profile for our survey sample. Over
40% of our survey respondents in Grade 1 reported ages
of between eight and nine. By Grade 6, just 12% of pupils
are at the right age-for-grade (11-years of age in this case).
These outcomes reflect a combination of delayed entry to
school, grade repetition and temporary drop out. Age-
for-grade improves at Grades 7 and 8, principally for the
perverse reason that many over-age children drop out.
12.1
8.5 7.7
6.4
3.4
…and no or low …and low schooling
education
teaching school fees costs
Boys
Child labour and education – a survey of slum settlements in Dhaka  55  
 Almost all of the working children in our survey have
some experience of education. Around 90% in the 11–14
age group report having attended school at some point.
These children typically started school later than their non-
working peers, pointing to delayed entry as a contributory
risk factor for early entry to work. They entered school at
an average age of 6.4, compared with 6 for non-working
children.
Working children enter the world of work having
accumulated limited years of school. Creating an education
profile for an average child worker is a hazardous exercise
– but it is one that can reveal underlying structures of
disadvantage. In the case of our survey, the average child
worker in the 11–14 age range dropped out of school at
the age of 10.5 having accumulated four years of school,
having completed a highest grade of 3.7 (Table 9). On an
uninterrupted progression pathway, children in the 11–14
age group would be expected to have completed between
Grades 5–8. The implication is that they started school
after the age of six and subsequently either repeated a
grade or dropped out.
The average inevitably obscures differences across ages.
Of the 8% of 10-year-olds who report being in work some
60% have not progressed beyond Grade 2. Almost one
third of 14-year-old working children report Grade 3 as the
highest level of attainment and almost three quarters ended
school at Grade 5 or less (Table 10). While our sampling
cannot be considered representative for individual age
groups, we reconfigured our survey respondent data to
capture the reported grade reached by working children at
the age they left school.

Figure 38. Misalignment in age-for-grade profiles: share of children by age in Grades 1–8

0.7 0.5
100 1.4
0.5
2.1 3.6
0.3 1.0
1.7 3.8 6.2
5.6
6.9 15.3
90 10.2
14.2 14.6
18.1 11.2
80 39.2
20.3
18.0
70 32.3 60.4
28.8 28.2
20.8
60
% 23.5
50
34.8
36.5
40 22.3
32.5 27.9 34.0
30 28.8
30.5
20
22.6 25.4 15.7
17.5 12.3
10 18.1
13.4 4.8
0.5 5.5 4.9 1.2 8.7
3.3 4.3
0 1.9 3.3

Grade 1 Grade 2 Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8

6 7 8 9 10 11 12 13 14

Table 9. The education profile of working children aged 11–14 years

Grade that the child

Current age of Age at which the child Age at which the child Average number of Highest completed
should have been
working child first enrolled in school left school years in school grade by age
when s/he left school
11 6.3 9.7 3.4 3.2 4
12 6.3 9.7 3.4 3.2 5
13 6.4 10.8 4.4 3.6 5
14 6.5 10.8 4.3 4.0 6
Age group 11–14 6.4 10.5 4.1 3.7 5

56  ODI Report

 The static snapshot taken at the age of 10–14 obscures
Table 10. Highest reported grade for working children
aged 10-14
the more complex moving picture of grade progression
by gender. Working girls are almost twice as likely as
Children aged 10 (%) Children aged 14 (%)
boys to reach Grade 5, though around one quarter report
Grades 1-3 as the highest level attained. Only 15% of Grade 1 22.3 5.6
working boys have competed Grade 5, with fully 38% Grade 2 38.8 12.4
only completing Grades 1-3. While there is a gender
Grade 3 27.8 14.7
advantage in grade completion for girls, the overall picture
is one of 14-year-old children entering labour markets Grade 4 - 18.9
with desperately low levels of education as measured by Grade 5 - 21.4
years of schooling. Factoring in the low levels of learning Grade 6 - 8.1
achievement, this reinforces the educational disadvantage
these children carry as they make the transition from the Grade 7 - 5.4
world of school to the world of work. Grade 8 - 4.5
The line separating working children from those out-of- Grade 9 - 0.6
school but not reported as being in employment is blurred.
Children reporting being ‘idle’ at the age of 11 may be
working by the age of 12 or 13. To explore the profile Grade repetition contributes to age-for-grade
of the group, we constructed an age-for-grade tracker misalignment. We did not observe marked differences
comparable to the exercise we carried out for working in repetition rates between children currently attending
children. As a group, ‘idle’ children have accumulated school, working children and children neither working
fewer years in school – just 3.7 on average. At the average nor attending school. However, the survey did detect some
age they left school they should have reached Grade 5, but marked differences in patterns of grade repetition. Children
had yet to complete Grade 3 (Table 11). currently working reported having lower repetition rates
for Grades 1–2, but far higher repetition rates in Grades
3–4 (Table 12).

Table 11. The education profile of non-working children who are out-of-school aged 11–14

Current age of child Grade that the child

Age at which the child Age at which the child Average number of Highest completed
neither working nor should have reached
first enrolled in school left school years in school grade by age
studying when s/he left school
11 6.7 9.6 2.9 2.7 4
12 6.7 9.9 3.2 3.0 5
13 6.8 10.8 4.0 3.4 5
14 6.9 11.7 4.8 4.1 6
Age group 11–14 6.8 10.5 3.7 3.4 5

Table 12. Patterns of grade repetition

Children that have dropped out of school

Children neither working nor
Children that are only studying (%) Working children (%)
studying (%)
Grade 1 39.4 34.3 27.4
Grade 2 23.9 29.0 25.0
Grade 3 13.4 10.7 23.8

Grade 4 4.9 6.6 12.9

Grade 5 1.0 7.4 3.0
Grade 6 1.2 0.0 1.5

Child labour and education – a survey of slum settlements in Dhaka  57  

 Several factors may contribute to these distinctive grade
repetition patterns. For children who entered school late,
the prospect of repetition at Grade 3 or Grade 4 may be
particularly damaging because it coincides with a typical
labour market entry age. Parents and, potentially, children
themselves, may view the opportunity cost of repetition as
being too high. It may also be the case that exam failure
is seen as signalling to parents and children that they are
unlikely to succeed in education.
We asked children to report on the reasons for grade
repetition. Here, too, there were some notable differences
(Figure 39). Currently working children were by far the
most likely to have repeated as a result of examination
failure, pointing to challenges in learning. For children
currently in school, and non-working out-of-school
children, migration and sickness also figured as significant
factors.
Why do children drop out of school? We asked that
question of out-of-school children and their parents.
What emerged was a broadly consistent pattern with
some marked differences (Figure 40). Working children
and their parents both cite economic pressure and school
costs as major factors. However, the children were far
more likely than their parents to see themselves as being
‘poor at studying’, suggesting they had internalised the
consequences of school failure. Conversely, one third of the
parents of working children cited truancy and the lack of
interest of the child in school as a reason for dropping out.
For non-working out-of-school children and their parents
the pattern is broadly similar. Children emphasise costs and
being ‘poor in studies’, while parents emphasise truancy
and costs.
Children report having attended different types of
school. Slums in Bangladesh are home to a range of
education providers, including the government, non-
governmental organisations (NGOs) (predominantly
not-for-profit) and private (for-profit) schools, and a range
of Madrashas. Working children aged 11–14 were far more
likely to have attended a government school or that of an
NGO provider, while private schools were the principal
provider for children attending school (Table 13).

Figure 39. Reasons for repetition: different groups of children attending and not attending school

80 72.3
70

60 52.5
50
% 39.5 39.3
40

30
20.7
20 12.7
6.6 8.7
10 2.7
0.6 0.0 1.5
0
Non-working children Working children
Children currently attending school Dropouts

Migration Failure in exam Sickness I had to work

Table 13. Schools attended by type and child status: in school, out-of-school and work, and child labourers, ages 11–14

What kind of school are you

School What kind of school did you last attend?
currently attending?
Non-working children that dropped Working children that dropped
Children only going to school (%)
out of school (‘idle’ children) (%) out of school (%)
Private school/college 43.4 18.7 15.2
Govt. school/college 27.1 51.6 55.7
NGO school 18.6 23.5 20.0
Madrasha 6.0 4.2 7.2

58  ODI Report

 Repetition rates were particularly high for children who Figure 41. Repetition rates for in-school and out-of-school
had dropped out of a government school, with over half children by school type – percentage of children repeating
reporting repetition of at least one grade (Figure 41). at least one grade
We asked parents of children attending school and
child labourers for the reasons behind their selection of 60
schools (in the case of working children, for the last school 52.5
attended). The quality of the school figured with more 50
prominence for parents in households with non-working
children (Figure 42). 40
31.6
There is a vibrant market for private tuition in Dhaka’s
slums – and this may be contributing to opportunity gaps % 30 24.7 23.4 23.4
between working and non-working children. Around half
20 15.0
of working children report having utilised private tutors
to improve exam results prior to dropping out of school. 10
Children currently attending school have higher rates of
utilisation (see Annexes). However, the biggest difference 0
is in levels of payment. We asked parents to report on Children current ly Children t hat dropped
the level of payments made to private tutors. Parents of at t ending school out of school
children currently attending school were more than twice
Govt. school/college
as likely to be spending in the range of Tk 500 to
Tk 1,000 a month, as parents of children who had dropped Non govt. school/college
out of school (Figure 43). To the extent that payment NGO school
levels serve as a proxy for hours of instruction and quality
of instruction, this could be an important transmission
mechanism for advantage and disadvantage in school.

Figure 40. Why do children drop out of school: children and parental perceptions by education and employment status

Main reasons that the child dropped out of school

100

90
8.0 6.9
14.1 3.7
80 4.9
32.6 8.8
70
19.3
60 32.7
20.3
5.1
% 50
21.7
12.2 6.4
40 26.7
11.2
30

20 26.6
19.7 35.1
28.4
10
8.2 5.7
0
Children Parents/guardian
Parents/guardians Children Parents/guardian
Parents/guardians
Working children Children neither working nor studying
The child needs to stay at home to look after siblings
School is too far from home
Truancy (the child does not want to go to school/does not consider education valuable)
The child is poor in studies
The child has to do paid work to earn money and supplement family income
Schooling costs (school fees, notebooks etc.) are too expensive
Other

Child labour and education – a survey of slum settlements in Dhaka  59  

 The small group of children combining work and Figure 43. Spending on private tuition varies across
school appears to struggle with a difficult balancing act. households: reported spending by band, differentiated by
On average, these children report working for just under education/employment status of child
three hours a day. By the time they are 11–14 years old,
they are half as likely as their non-working counterparts 100
in school to be attending at the correct grade for their 83.7 86.3
90
age (5.9% versus 11%) – albeit in a context where only 80
a minority are in the correct grade. Although absenteeism 70
rates are comparable for working and non-working 60 51.8
50

%
children attending school at around 13%, the reasons cited 38.4
40
for absenteeism vary. Over 80% of absenteeism among 30
children combining work and school is accounted for by a 16.3
20 12.3
7.9
reported need to work for payment or provide labour for 10 0.0 0.4
the family. 0
Non-­working   Working  
children children
3.4.2 What are children learning?
School attendance is a weak proxy for learning Children   Dropouts
achievement in Bangladesh, as in many other developing currently  
attending  
countries. As noted earlier, the marginal value of an school
additional year in school measured in terms of learning
outcomes is modest on average. In order to develop a Below  500  Takas
Below Tk 500Tk 500-­1,000  
Tk 500-1,000Takas
picture of learning achievement in our survey areas we
1,000-­2,000  
Tk 1,000-2,000TTk
akas
administered a series of simple literacy, numeracy and
reasoning tests. The results confirm a pattern of limited
average learning for all children, coupled with a marked
difference between working and non-working children.
Early entry into the world of work carries very significant
disadvantages in terms of the basic literacy and numeracy
skills that have a bearing on lifetime earnings and
livelihood security.

Figure 42. Reasons for school selection: parent (or guardian) responses on why they selected last school attended,
working children (last school attended) and non-working children

50 46.1
45 41.1
40
35
28.3
30
% 25 20.5 19.9
20 16.6
15 10.1 10.7
10
5 2.2 1.4 1.7 1.5
0
School is near Good No school Low tuition fees Other Siblings/children
home school/good fees/NGO’s from the
quality teaching school neighbourhood
are going to this
school

Households of non-working children Households of working children

60  ODI Report

 Our tests were designed to capture very basic levels of
learning achievement (see Annex 4 for technical details).
Difficulty levels were calibrated against competency
requirements expected of children in Grades 1–3. For
literacy, children were presented with 10 letters from the
Bengali alphabet and then five words in Bengali (Grades
1–2). They were then asked to read two simple sentences in
Bangla (Grade 3). These were as follows: ‘agricultural work
is hard’ and ‘the girl is playing’. Children were then asked
to read a simple passage in Bengali after which they were
tested for comprehension (Grade 3).
Our maths tests assessed children at levels of difficulty
spanning Grade 2–4. At the lower end of the range,
children were asked to order numbers and to identify
missing numbers in a sequence (Grades 2–3). The test
included standard Grade 3 level two-digit addition and
subtraction, and single digit multiplication. We also
included a subset of Grade 4–5 applied reasoning skills.
This included familiarity with formulae and use of
deduction. One of the questions was: ‘Aisha has 300 eggs.
She sells 218. What is the remaining number of eggs?’
All children were tested against the same criteria. We
divided respondents into two age groups, those aged 6–10
and those aged 11–14, and three comparison groups,
differentiating between child labourers, non-working
children and children neither in work nor in school. Tables

Table 14. Reading for basic literacy: correct identification of 10 letters in Bengali alphabet by age and work/education
status: share of children in age groups 6–10 and 11–14

Age group 6–10 Age group 11–14

Number of letters Non-working children Non-working children
that the child can Working Working
Children neither Children neither
correctly identify Children only children (%) Children only children
working nor working nor
studying (%) studying (%)
studying (%) studying (%)
0 0.8 17.1 22.2 0.5 11.1 7.3
1 2.1 15.4 8.9 0.7 8.0 2.0
2 1.3 9.0 11.1 0.2 5.0 3.2
3 2.3 12.2 13.3 0.4 4.0 4.1
4 1.4 6.9 4.4 0.3 4.2 1.7
5 3.2 10.1 6.7 0.6 6.1 4.6
6 2.3 5.3 2.2 0.6 4.4 3.2
7 3.7 3.2 11.1 0.7 1.7 3.3
8 6.7 3.7 4.5 2.2 9.8 5.7
9 10.9 6.4 2.2 3.0 10.0 7.3
10 65.2 10.6 13.3 90.8 35.6 57.7

Table 15. Reading for basic literacy: correct identification of five words in Bengali by age and work/education status:
share of children in age groups 6–10 and 11–14

Age group 6–10 Age group 11–14

Number of words Non-working children Non-working children
that the child can Working Working
Children neither Children neither
correctly read Children only children (%) Children only children (%)
working nor working nor
studying (%) studying (%)
studying (%) studying (%)
0 8.9 70.2 62.2 1.1 31.8 17.0
1 6.2 11.2 6.7 1.0 5.9 6.3
2 8.9 5.3 11.1 1.1 7.3 6.6
3 11.3 4.8 4.4 2.5 7.5 9.0
4 12.4 2.1 0.0 4.1 11.7 10.0
5 52.4 6.4 15.6 90.1 35.8 51.1

Child labour and education – a survey of slum settlements in Dhaka  61  

14 and 15 capture both the low level of basic literacy and
the divergence between children in school, on the one side,
and working and non-working out-of-school children on
the other. Among the indicative results:
•• For children who are only studying, one third of
6–10-year-olds are unable to correctly identify 10
Bengali letters and almost half are unable to read five
words correctly. Alarmingly, 10% of children aged
11–14 still in school were unable to achieve a wholly
correct score.
•• Children who enter labour markets between the ages
of 6–10 (around 14% of our survey sample) do so
with minimal literacy skills. Over half were unable to
correctly identify more than three Bengali letters and
almost two thirds were unable to read a single word
correctly.
•• Child labourers in aged 11–14 score very low on Grades
1–2 literacy tests. Only just over half were able to
identify the 10 Bengali letters and correctly read the five
simple words.
•• Only half of working children aged 11–14 were able to
read the sentence ‘the girl is playing’, compared to 89%
of children only studying (Figure 44).
•• Children who are neither working nor in school
perform even worse than working children, though this
finding has to be treated with some caution. In the case
of the 6–10 age group, many of these children have yet
to enter school.
These findings are disturbing on a number of levels. It
is evident that many working children learned very little
while attending school – and that the basic literacy skills
they did acquire erode very quickly. The performance of
children neither working nor studying is similarly a cause
for concern since a majority of these children are far more
likely to enter the world of work than re-enter the world of
school.
Raising the bar for learning achievement magnifies the
divergence between working children and those attending
school. Children aged 11–14 who participated in the
survey were asked to read a seven sentence passage (Table
16 and Annexes). The passage included simple information
on where a character lives, the number and age of her
siblings, and her ambitions for employment. The children
were tested for ability to read the passage, time taken and
comprehension. As highlighted in Table 16, while children
in school performed well below the levels anticipated in the
national curriculum, working children scored at far lower
levels. Fewer than half were able to read the passage in a
minute or less (compared to 87% for children in school).
Children neither working nor at school registered the
worst performance scores, though consideration has to be
given to the fact that this group includes some children
with delayed school entry.
Comprehension scores capture the very limited literacy
skills that working children accumulate, retain and take
with them into the workplace. Children aged 11–14 were
asked to answer a series of simple questions based on the
passage tested in Table 16. Working children on average
were half as likely as children in school to register correct
responses (Figure 45).
Maths test scores revealed differences as well as some
similarities with respect to test score. On basic Grade 3
numeracy, the gap between child labourers and children
attending school in the 11–14 age group was limited (Table
17). Working children are clearly able to build upon what
skills they leave school with in this area. This may reflect

Figure 44. Accurate reading of a basic literacy sentence: percentage of children correctly reading the sentence ‘the girl
is playing’ in Bengali

100 89.4 89.1

90
77.8
80
70
60 49.7 51.8
46.6
% 50
40 30.0 32.6 31.8
30 20.3 20.9
16.3
20 13.3
8.0 8.9 8.1
10 2.7 2.8
0
Children only Children neither Working children Children only Children neither Working children
studying working nor studying working nor
studying studying
Age group 6-10 Age group 11-14

Incorrect Partially correct Fully correct

62  ODI Report

the practical exigencies of mental arithmetic associated
with street vending, managing expenditure, supporting
families and negotiating wages. Children neither working
nor studying perform far below child labourers, suggesting
that the experience of work may be more important than
skills acquired at school.
One of the most striking results to emerge from the
survey is the desperately low level of applied maths
reasoning competency. Few of the 11–14-year-olds in
school are able to perform at Grade 3–4 level. Fully one
third was able to derive a result involving simple three-digit
subtraction; and just 17% could derive a multiplication
number. Child labourers performed very poorly on maths
reasoning skills, with just 5% able to correctly identify the
multiplier.
These learning results raise concerns at many levels.
For working children, the very low levels of basic literacy,
numeracy and applied learning capabilities is clearly a
constraint on prospects for employment and earnings.
These children have effectively been left with education
levels that will trap them in insecure, low-wage livelihoods.
The fact that basic literacy and numeracy skills acquired
in school have failed to ‘stick’, raises questions over the
quality of early grade teaching. More generally, if the slum-
dwelling children covered by our survey are in any sense
representative of children in other informal settlements
across Bangladesh, it is evident that the national education
system is failing to equip a large – and growing – section of
society with the skills and competencies needed to sustain
more dynamic and inclusive growth.

Table 16. Testing for reading: outcomes for children aged 6–10 and 11–14, proportion able to read and time taken

Pupils were asked to read the following passage in Bengali: My friend Nipa lives in Pabna. She loves that place. She has
two brothers and three sisters. Her brothers are older than her. My friend works at home every day. She takes care of her
two younger sisters at home. She wants to become a doctor after her studies.

Age group 6–10 Age group 11–14

Non-working children Non-working children
Children neither Working Children neither Working
Children only children (%) Children only children (%)
working nor working nor
studying (%) studying (%)
studying (%) studying (%)
The reader can read the passage in a 42.5 3.7 15.6 87.8 31.6 49.1
minute or less (% of children)
Time for reading the passage in seconds 55.5 58.6 57.8 44.1 56.6 53.4
The child can read the first five sentences 53.3 5.9 15.6 93.6 38.9 57.8
(% of children)

Figure 45. Testing for comprehension: share of children aged 11–14 correctly answering specified questions, by
education and employment status

100
87.5
83.4
77.6 78.9
65.1 65.5 68.0
61.8 61.6
51.5
46.5 47.1 44.1 48.0
% 50
37.6
26.0 23.0 29.3 25.7 28.3

0
Children only studying Children neither working
nor studying
Non-working children Working children Total for age group 11 to
14

Where does your friend live? (Answer: Pabna)

How many brothers and sisters does Nipa have? (Answer: Two brothers and three sisters)

Child labour and education – a survey of slum settlements in Dhaka  63  

3.4.3 Perceptions of school
How children perceive and experience school can have a
material bearing on what they are able to learn. On some
key indicators, our survey revealed divergent experiences
which may influence learning outcomes.
Child labourers and non-working but out-of-school
children report more negative experiences with respect
to their teachers than children currently attending school
(Table 18). They were less likely to feel they could express
an opinion or ask a question, and more likely to feel
mocked, teased or scolded. Child workers were twice as
likely to report having been scolded during their school
years. Children currently in school report greater levels
of confidence in asking teachers for advice (Figure 46).
Perhaps related to these experiences, children who have
dropped out, whether in work or out of work, report
greater difficulty in understanding textbooks.

Table 17. Numeracy and maths competencies: correct results by age group and education/employment status, children
aged 6–10 and 11–14

Age group 6–10 Age group 11–14

Non-working children Non-working children
Maths’ questions Working Working
Children neither Children neither
Children only children (%) Children only children (%)
working nor working nor
studying (%) studying (%)
studying (%) studying (%)
Arrange the numbers in descending order 85.0 39.4 71.1 97.7 76.8 88.9
8, 10, 5, 11, 4
Addition 51.7 10.6 35.6 90.5 32.6 64.6
37+57=?
Subtraction 58.6 12.2 48.9 93.4 43.5 75.2
18-7=?
Multiplication 45.7 4.8 26.7 86.7 31.4 51.2
6*7=?
Aisha has 300 eggs. She sells 218 eggs. 23.6 3.7 20.0 63.2 13.2 40.8
What is the remaining number of eggs that
Aisha is left with?
The result of the multiplication of two 3.5 0.0 0.0 17.2 0.0 4.8
numbers is 255. One of the number is 15.
What is the other number?

Table 18. Reported classroom experiences of children aged 6–14 by employment and education status: reporting on
perceptions of teachers

Children neither working

Children only studying (%) Working children (%)
nor studying (%)
In my class, I can/could express my opinion 86.0 70.6 77.0
In my class, I can/could ask questions to the teacher/s 90.2 77.4 83.8
How often do/did the teachers praise you for doing well in class?
a) Never 10.1 22.2 16.3
b) Some days 68.8 68.8 73.9
c) Every day 21.1 9.1 9.8
Do/did your teacher/s scold pupils when they do/did not understand the lesson?
a) No 57.2 50.5 43.4
b) Yes, some teachers 37.7 43.0 50.4
c) Yes, all the teachers 5.0 6.5 6.1

64  ODI Report

 Once again, considerable caution has to be exercised children living in informal settlements could do as well as
in interpreting these results. Home background and other children, with only 16% attributing failure to a lack
social characteristics of pupils will have had a bearing on of ability.
student learning outcomes. It may also be the case that
our survey is picking up changes over time: classroom Figure 47. What is the ability level of children living in
teaching practices may have improved since working slums and can they succeed in education: responses of
children dropped out. However, based on the reporting children aged 6–14 by education and employment status
from the children themselves it is difficult to avoid the
conclusion that negative attitudes on the part of teachers 90
interacting with other factors, may have fuelled a cycle of 77.0
80
low-expectation, poor learning outcomes and subsequent 68.5
drop-out. 70
Significant differences were reported with respect to the 60
physical environment by non-working children in school
50
as compared to child labourers (see Annex tables). The sole

%
exception to this rule related to an electricity connection 40
and the presence of an electric fan. 30
16.7
20 13.5
3.4.4 Parents’ and children’s expectations 10.3
10 1.0
Child workers appear to have internalised low levels of
expectation and to attribute limited education prospects 0
to personal failing (Figure 47). Over two thirds of those Childr en  in   Childr en  in   Childr en  in  
slums  have   low   slums  have   the   slums  can  do  as  
interviewed expressed the view that children living in ability same  ability  as   well  as  other  
informal settlements lack the ability to succeed at school, other s,   but  they   childr en  in  their  
str uggle   to  study studies
which is four times the share reported by non-working
children. Only 13% attributed difficulties in education Non-­working  children Working  children
to the economic pressures forcing them into work. By
contrast, three quarters of non-working children felt that

Figure 46. School experience of children and interaction of school with parents: child’s response to learning difficulties,
understanding of textbooks and parent-teacher meetings, by employment/education status of the child

90 83.1
75.6
80 70.9 73.0 70.5
70 63.2 60.2 57.1
%

60 50.2
50
40
30
20 14.8 15.7
7.3
10
0
Ask  t eacher Ask  another  s tudent
If  y ou  were/are  c onfused  by  a  c oncept  t aught  in  c lass,   The  child  f ound/finds   The  child's  s chool  
how  did/do  y ou  c larify  it? the  t extbooks   arranged/arranges  
easy  t o  understand regular  parent-­teacher  
meetings

Children  only  s tudying Children  neither  working  nor  s tudying Working  children

Child labour and education – a survey of slum settlements in Dhaka  65  

 By the time they are between the ages of 11–14, child
labourers and non-working children have very different
expectations, as do their parents (Figure 48). The majority
of working children see themselves working either in
‘business’ – in this context, usually street vending, running
a stall, or providing unskilled services – or in factories,
with a large proportion of girls anticipating a future as a
housewife. By contrast, non-working children anticipate a
future as a doctor, government worker, teacher or engineer.
This broadly mirrors the expectation of their parents
(Figure 48), who see education as a potentially secure route
towards public sector employment.
Being in work is associated with an elevated incidence
of negative feelings. Child labourers report lower
levels of confidence and happiness, and higher levels of
disappointment and frustration with themselves (Figures
49a, 49b and 49c). While the data has to be interpreted
with some caution, it suggests that children internalise
perceived self-failure as a set of outcomes over which
they have limited – if any – control. By extension, child
labourers also report lower levels of hope about the future.

Figure 48. Future expectations: what children and parents say about their future employment, by employment and
education status

40
33.9 33.7
35 31.4 31.8
30 27.7 27.2
25.6
25 20.5 20.6
18.9
20 16.2
%

16.8 13.6 13.9

15 9.2
12.7
9.5 9.9 10.3
7.1 6.9 7.9
10 4.6 4.8
6.6 5.1 5.6
5
0
Children Parents/guardians Children Parents/guardians Children Parents/guardians
Children  only  studying Working  children  (including  children   Children  neither   working  nor  studying
studying   and   working)

Businessman/woman Factory  worker Housewife Other  Government  J ob Doctor Teacher Engineer

66  ODI Report

Figure 49a. How often do you feel sad/unhappy?

Working children 54.2 38.1 5.8

Non-working children 72.1 24.8 2.3

0 10 20 30 40 50 60 70 80 90 100
%
Never Sometimes Most of the time Always

Figure 49b. My future is…

Working children 72.4 16.4 11.2

Non-working children 88.9 5.7 4.7

0 10 20 30 40 50 60 70 80 90 100
%
Bright Hopeles s Hopeles s and dis couraging

Figure 49c. How do you feel about yourself?

Working children 75.4 15.5 9.0

Non-working children 92.8 4.7 2.0

0 10 20 30 40 50 60 70 80 90 100
%
Confident and happy Sense of failure Source of my own troubles

Child labour and education – a survey of slum settlements in Dhaka  67  

Conclusions and policy
recommendations
This section briefly summarises some of the key policy
issues highlighted by the Child Labour Survey and
outlines our associated recommendations. Our hope is
that it will contribute to the broader national dialogue
on the elimination of (the worst forms) of child labour in
Bangladesh – and to the international dialogue on the role
of education in anti-child labour strategies.
Our survey findings serve to reinforce wider evidence on
the destructive power of the two-way interaction between
child labour and educational disadvantage. Children in
Dhaka’s slums enter labour markets for many reasons
and as a result of complex processes. Household poverty
and inherited social disadvantage – notably parental
education – figures prominently. The fact that earnings
from children’s work lifts the income of their households
to an average level comparable to that in households with
non-working children is important. It lends weight to the
proposition that parents are compelled to let their children
work in order to achieve a minimum level of income.
However, problems in the education system leading to high
drop-out rates during the primary education cycle is part
of the problem. It creates a ready supply of child entrants
to the labour market. The gap between Bangladesh’s
impressively high levels of primary school enrolment and
still depressingly low levels of completion are symptoms
of wider factors, including over-age enrolment, grade
repetition, and – critically – the quality of education.
Our survey findings highlight not just the scale of child
labour but also the severity of the problem. Almost all of
the working children in our sample are working much
more than 42 hours a week. The fact that many of these
children – especially girls – report working in the formal
garment sector raises questions about the effectiveness of
regulatory measures. For practical purposes, the reach of
regulatory institutions does not extend into the informal
sector. While the findings should be interpreted with
caution, our survey also points to employment practices –
notably cash payments with no contract – which add to the
vulnerability of working children.
Working children in Dhaka’s slums are overwhelmingly
out-of-school. Our survey does not capture the tensions
faced by children – young girls in particular – balancing
68  ODI Report
household chores with schooling. What it does capture is
the stark choice between education and employment that
working children make before completing the primary
cycle. That choice is the product of economic compulsion
and the educational failure generated by a schooling
system that under-serves slum-dwelling children in terms of
both access and quality.
In the absence of a concerted drive to eradicate the
worst forms of child labour, Bangladesh will not achieve
the 2030 development goal of universal secondary
schooling. Whatever the underlying drivers, the endemic
child labour in Dhaka’s slums constitutes a formidable
barrier to universal primary education. Developments in
these slums may or may not mirror experiences in other
urban centres. However, with around 30% of the children
covered in our survey either working or neither working
nor participating in school, the warnings signs are clear.
Rapid urbanisation in Bangladesh is not an automatic
route to accelerated progress in education. On the contrary,
in the absence of measures to expand opportunities for
schooling and to curtail child labour, it could be associated
with stagnation or even the reversal of past gains.
Breaking the link between child labour, social
disadvantage and restricted opportunities for education
will require a coordinated and comprehensive policy
response. There is currently a wide range of national
strategies, as well as bilateral and multilateral aid
interventions, aimed at tackling child labour. However,
the aggregate impact of the interventions now in place is
less than the sum of their parts. This is partly because of
problems in capacity and coordination across ministries;
and partly because the policy frameworks remain
fragmented. To take one obvious example: there is no
coherent education sector strategy spanning primary
and junior secondary education targeting universal
enrolment among slum-dwelling children. Coordination
between education ministries, the Ministry of Labour and
Employment and the Ministry of Women and Child Affairs
is also too weak to deliver results on the crisis, let alone at
the pace required.
Based on the findings set out in previous sections we
recommend policy reform in three key areas:
1 Building the evidence base
The Government of Bangladesh and aid partners need a
stronger evidence base on child labour and education in
urban slums. The Bangladesh Bureau of Statistics’ 2013
Child Labour Survey provides a high-quality nationally
representative sample survey covering over 1,500
enumeration areas. However, only 343 of these are in
urban locations – and there is no distinctive sample frame
for urban slums. Efforts should be made to link the next
Child Labour Survey to the updated Census of Slum Areas.
The 2014 Census provides a valuable breakdown of the
estimated distribution of slums and slum populations by
division. Outside of Dhaka, which accounts for half of the
total slum population, other urban centres with large and
expanding slum populations include Chittagong (27%
of the national total) Khulna (8%), Rangpur (5%) and
Sylhet and Rajshahi (both 4%). We recommend that the
Government of Bangladesh work with BRAC and others
to conduct bi-annual surveys of slums in each of these
divisions with a view to tracking levels of child labour
and the education status of working children. Survey
design should seek to capture the experience of children
themselves.
The Government of Bangladesh and international
partners should review national survey instruments to
improve the quality of information available. There is a
vast array of survey data available relating to child labour.
Beyond the national Child Labour Survey, sources range
from labour force surveys, to Census surveys, MICS and
household income and expenditure surveys. However, there
is little consistency in the design of survey instruments
and sampling frames. As part of the national strategy
for eliminating the worst forms of child labour, efforts
should be made to avoid duplication and to ensure that
each survey instrument adds a distinctive value to the
understanding of child labour patterns.
2 Making education a more effective part of
the strategy for eliminating child labour
Education must be placed at the centre of the national
strategy for combating child labour, with the age for
compulsory and free education provision raised from 10
to 14. There are a wide variety of national programmes
and projects in place aimed at expanding opportunities
for education. But the National Education Policy lacks
both a coherent strategy for preventing children aged 6-14
being drawn out of school and into labour markets – and
a strategy for getting children out of labour markets and
back into school. There should be a single integrated
strategy spanning the Ministry of Primary and Mass
Education and the Ministry of Education. Raising the age
for compulsory and free education to 14 would support
Bangladesh’s wider goals in education and child labour,
provided the right enabling environment is put in place to
counteract the pressures forcing children to enter labour
markets.
Bangladesh is underinvesting in education. Despite the
increase in primary and secondary school enrolment over
the past 15 years, Bangladesh continues to spend less than
3% of GDP (and 14–16% of the budget) on education.
This is well below the levels recorded in sub-Saharan
Africa and the average for low-income countries. An
additional concern is that the capital budget in education
is skewed against the most disadvantaged areas. The
underfinancing of education is reflected in the limited
availability of public school provision in slums, classroom
overcrowding and a range of education quality concerns.
While there is scope for enhanced efficiency in education
spending, expanded financial resources and greater
equity in resource allocation have a critical role to play in
expanding opportunities for education in slum areas. As an
indicative target, Bangladesh should aim to spend 4–5% of
national income on education by 2020.
Demand-side financing and cash transfers could play an
expanded role in strengthening school participation among
urban slum-dwelling children. The parents of working
children in slums and the children themselves are typically
forced to choose employment over education because
of economic necessity. The unaffordability of the direct
costs of schooling (and, increasingly, private tuition fees)
and the indirect costs, or opportunity costs, of foregoing
child labour are too high. Cash transfers can lower these
costs and expand choice. Bangladesh has pioneered
some of the world’s largest cash transfer programmes in
education, ranging from the Primary Education Stipend
Programme, stipends for girl students and a range of
targeted programmes. Taken collectively, however, these
programmes attach insufficient weight to the circumstances
of slum-dwelling children forced into labour markets.
Transfers at the primary level are too low to compensate
working children and their parents for the opportunity
costs of staying in school. Indeed, slum-dwelling children
have been largely bypassed. Those dropping out of school
do not benefit from the primary education stipends or the
girl student secondary education stipends. Consideration
should be given to the creation of large-scale targeted cash
transfer programmes for children living in slums, with an
emphasis on front-loaded support during the early grades.
Both conditional and unconditional cash transfers may
have a role to play. Current cash transfers for education
are conditional on school attendance – and evidence from
other countries points to the success of such transfers in
curtailing child labour. However, unconditional transfers
targeting marginalised families may also have a role to play
in reducing the economic pressures driving children into
the work place. The impact of any transfer programme
will be determined, in part, by the level of the transfer. Our
survey documents average monthly wages for 6–14-year-
old child labourers in Dhaka’s slums of Tk 3,664, around
$47 – and this may provide an indication of the level of
Child labour and education – a survey of slum settlements in Dhaka  69  
support needed to shift household incentives. Aligning
the level and timing of transfers with the critical risk
factors associated with school drop-out (in particular,
late enrolment and grade repetition) could increase the
efficiency of transfers. However, interventions in education
cannot be considered in isolation. Education outcomes
will be determined in part by the degree to which the most
vulnerable households are insulated through wider social
protection measures from the exogenous shocks associated
with food price increases, flooding and other events that
increase poverty.
Urgent action is needed to improve the learning
environment. Children living in slums experience, in acute
form, the wider education quality problems evident in
Bangladesh – and many carry the consequences of an
abysmal learning experience with them into the world
of work. The very low levels of learning reported in the
early grades and the limited value of an additional year of
schooling point to systemic problems. As highlighted in
previous sections, there is an urgent need to improve the
quality of teaching, and teachers should be better-trained
and supported to help first generation learners. At the
same time, slum-dwelling children carry with them into
the classroom the disadvantages that come with poverty
and non-literate home environments. Evidence from a
range of developing countries suggests that early childhood
programmes can be highly effective in addressing these
disadvantages – and there are compelling grounds for the
development of a national programme to deliver universal
early childhood programmes in slum areas.
3 Linking education to strengthened child
labour legislation and a more robust
regulatory environment
Bangladesh could further strengthen the legislative
framework for combating child labour. Recent years
have seen the Government of Bangladesh adopt an
increasingly robust framework for combating child labour.
The Children’s Act (2013), the National Children Policy
(2011) and the National Plan of Action (2010) establish a
set of well-defined and ambitious goals, and identify lead
ministries and cross-ministry responsibilities. However, the
current framework suffers from a number of weaknesses.
The ‘42-hour’ threshold for child labour (as distinct from
child work) is excessive and the gazetted list of hazardous
work remains partial. The conclusion drawn by an earlier
report remains valid: ‘current legislation still does not
constitute a comprehensive legal framework for protecting
children against child labour, and in particular those
working in the informal sector’ (UCW, 2011). Ratification
of ILO Convention 138 concerning the minimum age
70  ODI Report
for admission to employment is overdue. Moreover, the
machinery for enforcement and monitoring located in the
Office of the Chief Inspector of Factories in the Ministry
of Labour and Employment is inadequately resourced.
Cases of sanction applied to employers are few and far
between – and fines are pitched at a derisory level. Backing
stronger legislation with a strengthened regulatory capacity
and more punitive sanctions is critical. Beyond investing
in human resources in the Ministry of Labour and
Employment, social work agencies and the police, building
a more comprehensive child monitoring system will require
the deepening of international partnerships with agencies
such as the ILO and UNICEF, and engagement with
NGOs.
The two ministries of education should be coordinating
more effectively with the Ministry of Labour and
Employment and other agencies. Child work that keeps
children out of school is a defining feature of child
labour – and our survey suggests that most child work in
slums is conducted by out-of-school children. An obvious
corollary is that schools should be a key reporting conduit
on child labour. School authorities should be engaging with
community-based groups and municipal authorities to
identify children who are either working or at risk of child
labour as a consequence of school drop-out.
The Dhaka City Corporation (DCC) could set a new
national standard for enforcement. The DCC could bring
together employers, school authorities, community groups
and slum-dweller associations to provide more effective
monitoring. It could also integrate child labour standards
into its trade licensing systems, which are overseen by tax
officers.
The formal garment sector needs strengthened
regulatory oversight. While emphasising that our survey
results are based on reporting by children rather than
work place observation, they point to a high level of child
labour in the formal garment sector. This is a matter of
concern on many levels. Child labour in the garment
sector cuts against the aims of national education policy.
It also poses a potential threat to export markets and
foreign investment. There are compelling grounds for
the Government of Bangladesh, employer associations
and community groups to come together in addressing
the problem – and this should be a priority for the Chief
Inspector of Factories.
Accelerated progress towards universal birth registration
is vital. One of the difficulties facing employers can be
traced to the low rate of birth registration in Bangladesh’s
slums, which in turn makes age verification difficult.
It follows that any strengthening of the regulatory
environment is contingent on an accelerated drive towards
universal birth registration.
References
Ahmed, I. (2014). ‘Factors in Building Resilience in Urban Slums of Dhaka, Bangladesh’, Procedia Economics and
Finance 18: 745-753.
Ahmed, F.; Ahmad, N.; Khan, M.A.; Jolliffe, D.; Mahbub, M.A.; Sharif, l.; Yoshida, N.; Zaidi, S.; Swaroop, V. and Zutt,
J. (2010) Poverty Maps of Bangladesh 2010: Key findings. Washington, DC: World Bank. Available online: http://
documents.worldbank.org/curated/en/916761468211763695/pdf/904870WP0WB0Po00Box385319B00PUBLIC0.pdf.
Akharuzzaman, M. (2014) ‘Urbanization and Environmental Concept of Dhaka City, Bangladesh’. Fukuoka: Fukuoka
Asian Urban Research Center (URC). Available online: http://urc.or.jp/wp-content/uploads/2014/06/0612_bangla_2.
pdf.
Al-Zayed, S.R. (2016) Bangladesh - BD: Reaching Out of School Children II : P131394 - Implementation Status Results
Report: Sequence 07. Washington, DC: World Bank. Available online: http://documents.worldbank.org/curated/
en/696621471580542033/pdf/ISR-Disclosable-P131394-08-19-2016-1471580527273.pdf.
Alvi, E. and Dendir, S. (2011) ‘Weathering the Storms: Credit receipt and child labor in the aftermath of the great floods
(1998) in Bangladesh’, World Development 39(8): 1398-1409.
Angeles, G.; Lance, P.; Barden-O’Fallon, J.; Islam, N.; Mahbub, A.Q.M. and Nazem, N.I. (2009) ‘The 2005 Census and
Mapping of Slums in Bangladesh: Design, select results and application’, International Journal of Health Geographics
8(1): article 32.
Asadullah, M.N. and Chaudhury, N. (2013) Primary Schooling, Student Learning, and School Quality in Rural
Bangladesh. Center for Global Development. Working Paper 349. Washington, DC: Center for Global Development.
Available online: http://www.cgdev.org/sites/default/files/primary-schooling-student-learning-school-quality-rural-
bangladesh_1.pdf.
Baird, S., McIntosh, C. and Özler, B. (2011) ‘Cash or Condition? Evidence from a cash transfer experiment’, Quarterly
Journal of Economics 26(4): 1709-1753.
Baland, J.-M. and Robinson, J.A. (2000) ‘Is Child Labor Inefficient?’, Journal of Political Economy 108(4): 663-679.
BANBEIS and Bangladesh Ministry of Education (2015) Bangladesh Education Statistics – 2015. Dhaka: Bangladesh
Bureau of Educational Information and Statistics (BANBEIS).
Bangladesh Bureau of Statistics (2015) Preliminary Report on Census of Slum Areas and Floating Population 2014.
Dhaka: Bangladesh Bureau of Statistics. Available online: http://www.bbs.gov.bd/WebTestApplication/userfiles/Image/
Slum/Preli_Slum_Census.pdf.
Bangladesh Bureau of Statistics (2013) Child Labour Survey Bangladesh 2013. Dhaka: Bangladesh Bureau of Statistics
and International Labour Organization. Available online: http://www.bbs.gov.bd/WebTestApplication/userfiles/Image/
LatestReports/ChildLabourSurvey2013.pdf.
Bangladesh Bureau of Statistics and UNICEF Bangladesh (2014) Progotir Pathey Multiple Indicator Cluster Survey 2012-
2013: Key findings. Dhaka: Bangladesh Bureau of Statistics and UNICEF Bangladesh. Available online: http://www.
unicef.org/bangladesh/MICS_2012-2013__Key_Findings.pdf.
Bangladesh Ministry of Education (2010) National Education Policy 2010. Dhaka: Government of the People’s Republic
of Bangladesh.
Barham, T., Macours, K. and Maluccio, J.A. (2013) More Schooling and More Learning? Effects of a three-year
conditional cash transfer program in Nicaragua after 10 Years. IDB Working Paper Series No. IDB-WP-432.
Washington, DC: Inter-American Development Bank. Available online: https://publications.iadb.org/bitstream/
handle/11319/4584/MORE%20SCHooling%20and%20more%20learning%20full%20body.pdf?sequence=1 .
Bastagli, F.; Hagen-Zanker, J.; Harman, L.; Sturge, G.; Barca, V.; Schmidt, T. and Pellerano, L. (2016) Cash Transfers:
What Does the Evidence Say? A rigorous review of impacts and the role of design and implementation features.
London: Overseas Development Institute. Available online: https://www.odi.org/publications/10505-cash-transfers-
what-does-evidence-say-rigorous-review-impacts-and-role-design-and-implementation.
Baker, J. L. (2007) Dhaka: improving living conditions for the urban poor. Bangladesh Development Series. World Bank
Basu, K. (1999) ‘Child Labor: Cause, consequence, and cure, with remarks on international labor standards’, Journal of
Economic Literature 37(3): 1083-1119.
Basu, K. and Van, P.H. (1998) ‘The Economics of Child Labor’, The American Economic Review 88(3): 412-427.
Becker, G. (1964) Human Capital: A theoretical and empirical analysis, with special reference to education. New York:
Columbia University Press.
Behrman, J.R., Parker, S.W. and Todd, P.E. (2011) Do Conditional Cash Transfers for Schooling Generate Lasting
Benefits?: A Five-Year Follow-up of PROGRESA/Oportunidades’, Journal of Human Resources, 46(1): 93-122.

Child labour and education – a survey of slum settlements in Dhaka  71  

Behrman, J., Parker, S.W. and Todd, P.E. (2009) ‘Schooling Impacts of Conditional Cash Transfers on Young Children:
Evidence from Mexico’, Economic Development and Cultural Change 57(3): 439-477.
Bharadwaj, P., Lakdawala, L.K. and Li, N. (2013) Perverse Consequences of Well-intentioned Regulation: Evidence from
India’s Child Labor Ban. NBER Working Paper No. 19602. Cambridge, MA: National Bureau of Economic Research.
Available online: http://www.nber.org/papers/w19602.
Brewer, J. and Porter, R. (1994) Consumption and the World of Goods. Abingdon: Routledge.
Cameron, S. (2011) ‘Whether and Where to Enrol? Choosing a primary school in the slums of urban Dhaka, Bangladesh’,
International Journal of Educational Development 31(4): 357-366.
Cameron, S. (2010) Access to and Exclusion from Primary Education in Slums of Dhaka, Bangladesh. Create Pathways
to Access Research Monograph No. 45. Falmer: Consortium for Educational Access, Transitions and Equity
(CREATE). Available online: http://www.create-rpc.org/pdf_documents/PTA45.pdf.
Cameron, S. (2009) ‘Education in Slums of Dhaka, Bangladesh’. Paper prepared for the 10th UKFIET International
Conference on Education and Development. September 15–17. Available online: http://www.create-rpc.org/
pdf_documents/UKFIETstuartcameronpaper.pdf.
Cox, W. (2012) ‘Evolving Urban Form: Dhaka’. NewGeography.com. Available online: http://www.newgeography.com/
content/003004-evolving-urban-form-dhaka.
Cross Validated (2016) ‘How Do I Interpret a Probit Model In Stata?’ Cross Validated. Available online: http://stats.
stackexchange.com/questions/42956/how-do-i-interpret-a-probit-model-in-stata
Cunningham, H. and Viazzo, P.P. (eds) (1996) Child Labour in Historical Perspective, 1800-1985: Case studies from
Europe, Japan and Colombia. Florence: UNICEF.
de Hoop, J.; Friedman, J.; Kandpal, E. and Rosati, F.C. (2016) Complementarities Between Schooling and Child Work in
the Presence of an Education Subsidy. Rome: Understanding Children’s Work (UCW).
de Hoop, J. and Rosati, F.C. (2014) Cash Transfers and Child Labor. World Bank Policy Research Working
Paper No. 6826. Washington, DC: World Bank. Available online: http://documents.worldbank.org/curated/
en/749621468338433968/Cash-transfers-and-child-labor.
Del Carpio, X.V. and Loayza, N. (2012) The Impact of Wealth on the Amount and Quality of Child Labor. World Bank
Policy Research Working Paper No. 5959. Washington, DC: World Bank. Available online: https://openknowledge.
worldbank.org/handle/10986/9.
DHS (2016a) ‘Wealth Index Construction’. Rockville, MD: The Demographic and Health Surveys (DHS) Program.
Available Online: http://dhsprogram.com/topics/wealth-index/Wealth-Index-Construction.cfm.
DHS (2016b) ‘What is the Wealth Index?’. Rockville, MD: The Demographic and Health Surveys (DHS) Program.
Available online: http://www.dhsprogram.com/topics/wealth-index/Index.cfm.
Diallo, Y., Etienne, A. and Mehran, F. (2013) Global Child Labour Trends 2008 to 2012. Geneva: International Labour
Office. Available online: http://www.ilo.org/ipec/Informationresources/WCMS_IPEC_PUB_23015/lang--en/index.htm.
Dobbs, R.; Smit, S.; Remes, J.; Manyika, J.; Roxburgh, C. and Restrepo, A. (2011) Urban World: Mapping the economic
power of cities. New York: McKinsey Global Institute. Available online: http://www.mckinsey.com/global-themes/
urbanization/urban-world-mapping-the-economic-power-of-cities.
Edmonds, E.V. (2009) Defining Child Labour: A review of the definitions of child labour in policy research. Geneva:
International Labour Office, International Programme on the Elimination of Child Labour (IPEC). Available online:
http://www.ilo.org/ipecinfo/product/viewProduct.do?productId=11247.
Fasih, T. (2007) Analyzing the Impact of Legislation on Child Labor in Pakistan. Policy Research Working Paper No.
4399. Washington, DC: World Bank.
FREPD (Foundation for Research on Educational Planning and Development) (2003) Baseline Survey of Street Children
in Bangladesh - Report. Dhaka : Bangladesh Bureau of Statistics. Available online: www.ilo.org/ipecinfo/product/
download.do?type=document&id=287.
Fry, K., Firestone, R. and Chakraborty, N.M. (2014) Measuring Equity with Nationally Representative Wealth Quintiles.
Washington, DC: PSI. Available online: http://www.psi.org/wp-content/uploads/2014/10/Wealth-Quintile-Guide.pdf.
Galiani, S. and McEwan, P.J. (2013) ‘The Heterogeneous Impact of Conditional Cash Transfers’, Journal of Public
Economics 103(July): 85–96.
Gitter, S.R. and Barham, B.L. (2008) ‘Women’s Power, Conditional Cash Transfers, and Schooling in Nicaragua’, World
Bank Economic Review 22(2): 271-290.
Glewwe, P. and Kassouf, A.L. (2012) ‘The Impact of the Bolsa Escola/Família Conditional Cash Transfer Program on
Enrollment, Dropout Rates and Grade Promotion in Brazil’, Journal of Development Economics 97(2): 505-517.
Glewwe, P. and Muralidharan, K. (2015) Improving School Education Outcomes in Developing Countries: Evidence,
Knowledge Gaps, and Policy Implications. RISE Working Paper-WP-15/001. Oxford: RISE – Research on Improving

72  ODI Report

 Systems of Education. Available online: http://www.bsg.ox.ac.uk/sites/www.bsg.ox.ac.uk/files/documents/RISE_
WP-001_Glewwe_Muralidharan.pdf.
Goldin, C. and Katz, L. (2003) Mass Secondary Schooling and the State: The role of state compulsion in the high school
movement. NBER Working Paper 10075. Available online: http://www.nber.org/papers/w10075.pdf.
Government of the People’s Republic of Bangladesh (2013) Bangladesh Primary Education Stipends: A qualitative
assessment. Dhaka: Directorate of Primary Education, Power and Participation Research Centre and UNICEF
Bangladesh. Available online: http://www.unicef.org/bangladesh/Bngladesh_Primary_Education_Stipends_survey.pdf.
Government of the People’s Republic of Bangladesh (2014) Bangladesh Primary Education Annual Sector Performance
Report – 2014. Dhaka: Directorate of Primary Education & Monitoring and Evaluation Division. Available
online: http://dpe.portal.gov.bd/sites/default/files/files/dpe.portal.gov.bd/publications/e59c1e65_d871_446e_
a5f5_3ab69fee590b/Final%20ASPR%202014.pdf.
Government of the People’s Republic of Bangladesh (2006) Labour Act, 2006 (XLII of 2006). Dhaka: Government of the
People’s Republic of Bangladesh. Available online: http://www.dpp.gov.bd/upload_file/gazettes/14212_75510.pdf.
Grootaert, C. and Kanbur, R. (1995) Child Labor: A Review. World Bank Policy Research Working Paper No. 1454.
Washington, DC, World Bank. Available online: http://ssrn.com/abstract=620526.
Gruebner, O.; Sachs, J.; Nockert, A.; Frings, M.; Khan, M.M.H.; Lakes, T. and Hostert, P. (2014) ‘Mapping the Slums of
Dhaka from 2006 to 2010’, Dataset Papers in Science 7: Article ID 172182.
Guarcello, L., Lyon, S. and Valdivia, C. (2015) Evolution of the relationship between child labour and education
since 2000. Evidence from 19 developing countries. Understanding Children’s Work Programme Working
Paper Series. Rome: Understanding Children’s Work (UCW). Available online: http://unesdoc.unesco.org/
images/0023/002324/232455e.pdf.
Hopkins, E. (1994) Childhood Transformed: Working-class children in nineteenth century England. Manchester:
Manchester University Press.
Horrell, S. and Humphries, J. (1995) ‘”Exploitation of Little Children”: Child labor and the family economy in the
Industrial Revolution’, Exploration in Economic History 32(4): 485-516.
Humphries, J. (2012) ‘Childhood and Child Labour in the British Industrial Revolution’, Economic History Review
66(2): 395–418.
ILO (2012) International Standard Classification of Occupations: Structure, group definitions and correspondence tables
(ISCO-08) Volume 1. Geneva: International Labour Office. Available online: http://www.ilo.org/wcmsp5/groups/
public/---dgreports/---dcomm/---publ/documents/publication/wcms_172572.pdf.
ILO (2006) Baseline Survey on Child Domestic Labour (CDL) in Bangladesh. Dhaka: International Labour Office.
Available online: http://www.ilo.org/ipec/Informationresources/WCMS_IPEC_PUB_4647/lang--en/index.htm
ILO (1999) Convention 182 - C182 - Worst Forms of Child Labour Convention, 1999 (No. 182) Convention concerning
the Prohibition and Immediate Action for the Elimination of the Worst Forms of Child Labour (Entry into force:
19 Nov 2000). Adoption: Geneva, 87th ILC session (17 Jun 1999) - Status: Up-to-date instrument (Fundamental
Convention). Geneva: International Labour Organization. Available online: http://www.ilo.org/dyn/normlex/
en/f?p=1000:12100:0::NO::P12100_ILO_CODE:C182.
ILO (1973) Convention 138 - C138 - Minimum Age Convention, 1973 (No. 138) - Convention concerning Minimum
Age for Admission to Employment (Entry into force: 19 Jun 1976). Adoption: Geneva, 58th ILC session (26 Jun 1973)
- Status: Up-to-date instrument (Fundamental Convention). Geneva: International Labour Organization. Available
online: http://www.ilo.org/dyn/normlex/en/f?p=NORMLEXPUB:12100:0::NO::P12100_ILO_CODE:C138.
ILO, Governance and Tripartism Department, and International Programme on the Elimination of Child Labour (2013)
Marking Progress Against Child Labour: Global estimates and trends 2000-2012. Geneva: International Labour
Office. Available online: http://www.ilo.org/wcmsp5/groups/public/---ed_norm/---ipec/documents/publication/
wcms_221513.pdf.
INCIDIN Bangladesh (2006) Aggregate Report on Baseline and Situation Analysis for Child Labour Projects of SCSD
Partner’s Project Area. Stockholm: Save the Children Sweden-Denmark.
Innovations for Poverty Action (2011) ‘Evaluations of Cash Transfer Programs in African Settings’. Policy memo. New
Haven, CT: Innovations for Poverty Action. Available online: https://www.poverty-action.org/sites/default/files/
publications/evaluations_of_cash_transfer_programs_in_african_settings_policy_memo.pdf.
Islam, M.S.; Anannya, A.M.; Rahman, M.S. and Rahman, M.M. (2015) ‘Present Situation of Water Supply and Sanitation
at Karail Slum Dhaka’, Journal of Environmental Science and Natural Resources 8(1): 161-163.
Islam, N.; Mahbub, A.; Nazem, N.I.; Angeles, G. and Lance, P. (2006) Slums of Urban Bangladesh: Mapping and Census,
2005. Dhaka: Centre for Urban Studies, MEASURE Evaluation, National Institute of Population Research and
Training. Available online: https://www.measureevaluation.org/resources/publications/tr-06-35.

Child labour and education – a survey of slum settlements in Dhaka  73  

Jolliffe, D.; Sharif, I.; Gimenez, L. and Ahmed, F. (2013) Bangladesh – Poverty Assessment: Assessing
a decade of progress in reducing poverty for the period 2000-2010. Washington, DC: World
Bank. Available online: http://documents.worldbank.org/curated/en/370861468003615077/
Bangladesh-Poverty-assessment-assessing-a-decade-of-progress-in-reducing-poverty-for-the-period-2000-2010.
Khan, S. and Lyon, S. (2015) Measuring Children’s Work in South Asia: Perspectives from national household surveys.
New Delhi: International Labour Organization; ILO DWT for South Asia and ILO Country Office for India. Available
online: http://www.ilo.org/wcmsp5/groups/public/---asia/---ro-bangkok/---sro-new_delhi/documents/publication/
wcms_359371.pdf.
Khandker, S., Pitt, M. and Fuwa, N. (2013) Subsidy to Promote Girls Secondary Education: The Female Stipend
Program in Bangladesh. enGender Impact: the World Bank’s Gender Impact Evaluation Database. Washington
DC: World Bank. Available online: http://documents.worldbank.org/curated/en/183331468013214760/
Subsidy-to-promote-girls-secondary-education-the-female-stipend-program-in-Bangladesh.
Labowitz, S. and Baumann-Pauly, D. (2015) Beyond the Tip of the Iceberg: Bangladesh’s Forgotten Apparel
Workers. Center for Business and Human Rights, NYU STERN. Available online: http://static1.squarespace.com/
static/547df270e4b0ba184dfc490e/t/5672bef1c647ad862c494795/1450360561022/Bangladesh-Report-Final.pdf
Lucci, P. and Bhatkal, T. (2014) Monitoring Progress on Urban Poverty: Are current data and indicators fit for purpose?
ODI Working Paper 405. London: Overseas Development Institute. Available online: https://www.odi.org/sites/odi.org.
uk/files/odi-assets/publications-opinion-files/9206.pdf.
Ministry of Labour and Employment (2013). National Plan of Action for Implementing the National Child Labour
Elimination Policy 2012-2016. Dhaka: Government of the People’s Republic of Bangladesh.
Mohit, M.A. (2012) ‘Bastee Settlements of Dhaka City, Bangladesh: A review of policy approaches and challenges ahead’,
Procedia - Social and Behavioral Sciences 36: 611-622.
MoPME (2014) National Student Assessment 2013 for Grades 3 and 5. Ministry of Primary & Mass Education,
Monitoring and Evaluation Division, Directorate of Primary Education. Available online: http://mopme.portal.gov.bd/
sites/default/files/files/mopme.portal.gov.bd/page/485eacef_5746_4c37_b88f_f01f0a1250b7/NSAFinal%202013%20
Report%2019012015.compressed.pdf.
NIPORT, Mitra and Associates and ICF International (2016) Bangladesh Demographic and Health Survey 2014.
Dhaka and Rockville, MD: National Institute of Population Research and Training, Mitra and Associates and ICF
International. Available online: https://dhsprogram.com/pubs/pdf/FR311/FR311.pdf.
O’Halloran, S. (nd) Sustainable Development U9611 Econometrics II - Lecture 9: Logit/Probit [Online]. Available: http://
www.columbia.edu/~so33/SusDev/Lecture_9.pdf.
Parajuli, D. (2016) Bangladesh - Secondary Education Quality and Access Improvement: P106161 - Implementation
Status Results Report: Sequence 15. Washington, DC : World Bank. Available online: http://documents.worldbank.org/
curated/en/155271468014378053/Bangladesh-Bangladesh-Secondary-Education-Quality-and-Access-Improvement-
P106161-Implementation-Status-Results-Report-Sequence-15.
Piza, C. and Portela Souza, A. (2016) Short- and Long-term Effects of a Child-Labor Ban. Policy Research Working
Paper No. WPS 7796, Impact Evaluation series. Available online: http://documents.worldbank.org/curated/
en/146211471281195366/pdf/WPS7796.pdf.
Rosenzweig, M.R. and Evenson, R. (1977) ‘Fertility, Schooling, and the Economic Contribution of Children in Rural
India: An econometric analysis’, Econometrica 45(5): 1065-1079.
Rutstein, S.O. and Johnson, K. (2004) The DHS Wealth Index. DHS Comparative Reports No. 6. Calverton: ORC
Macro. Available online: https://dhsprogram.com/pubs/pdf/CR6/CR6.pdf.
Sabates, R., Hossain, A. and Lewin, K.M.I. (2010) School Drop Out in Bangladesh: New Insights from Longitudinal
Evidence. CREATE Pathways to Access Series. Research Monograph Number 49. Falmer: Consortium for Educational
Access, Transitions and Equity (CREATE). Available online: http://www.create-rpc.org/pdf_documents/PTA49.pdf.
Schurmann, A.T. (2009) ‘Review of the Bangladesh Female Secondary School Stipend Project Using a Social Exclusion
Framework’, Journal of Health, Population, and Nutrition 27(4): 505-517.
Sen, A. (1999) Development as Freedom. Oxford: Oxford University Press.
Shafi, S.A. and Payne, G. (2007) ‘Local Partnerships for Urban Poverty Alleviation: Land Tenure Security and Land
Administration’. Dhaka: Final Report to LGED, UNDP and UN-Habitat.
Soares, J. (2003) ‘The Economics of Child Labor: Another comment’. Unpublished paper. Department of Economics,
George Washington University. Available online: http://udel.edu/~jsoares/basuvan.pdf.
Steer, L., Rabbani, F. and Parker, A. (2014) Primary Education Finance for Equity and Quality: An analysis of past
success and future options in Bangladesh. Brooke Shearer Working Paper Series, Working Paper 3. Washington,
DC: The Brookings Institution. Available online: https://www.brookings.edu/wp-content/uploads/2016/06/Equity-
Financing-in-Bangladesh-Final-v2.pdf.

74  ODI Report

UCW (2015) The Twin Challenges of Child Labour and Educational Marginalisation in the East and South-East Asia
Region: Preparing for a post 2015 world. Rome: Understanding Children’s Work Programme. Available online: http://
www.ilo.org/ipec/Informationresources/WCMS_IPEC_PUB_25520/lang--en/index.htm.
UCW (2011) Understanding Children’s Work in Bangladesh Country Report 2011. Rome: Understanding
Children’s Work Programme. Available online: http://ucw-project.org/attachment/Bangladesh_child_labour_
report20111125_094656.pdf.
UNDP (2015) ‘Briefing note for countries on the 2015 Human Development Report – Bangladesh’. New York: United
Nations Development Programme. Available online: http://hdr.undp.org/sites/all/themes/hdr_theme/country-notes/
BGD.pdf.
UNESCO World Inequality Database on Education (WIDE) - Data on Bangladesh. Available online: http://www.
education-inequalities.org/countries/bangladesh#?dimension=all&group=all&year=latest.
UNICEF (2014) All Children in School by 2015 Global Initiative on Out-of-School Children: South Asia Regional Study
Covering Bangladesh, India, Pakistan and Sri Lanka. Nepal: UNICEF Regional Office for South Asia. Available online:
http://www.unicef.org/education/files/SouthAsia_OOSCI_Study__Executive_Summary_26Jan_14Final.pdf.
United Nations (2016) Transforming our world: The 2030 Agenda for Sustainable Development. New York: United
Nations. Available online: https://sustainabledevelopment.un.org/post2015/transformingourworld.
United Nations General Assembly (2015) Resolution adopted by the General Assembly on 25 September 2015. New
York: United Nations. Available online: http://www.un.org/ga/search/view_doc.asp?symbol=A/RES/70/1&Lang=E.
United States Department of Labor (2013) Findings on the Worst Forms of Child Labor: Bangladesh moderate
advancement. Washington, DC: Bureau Of International Labor Affairs. Available: https://www.dol.gov/ilab/reports/
child-labor/findings/2013TDA/bangladesh.pdf.
USAID and DFID (2014) Inclusive Growth Diagnostic Bangladesh. Washington, DC and London: US Agency for
International Development and UK Department for International Development.
Williams, R. (2016) Using Stata’s Margins Command to Estimate and Interpret Adjusted Predictions and Marginal
Effects. Notre Dame: University of Notre Dame. Available online: https://www3.nd.edu/~rwilliam/stats/Margins01.pdf
World Bank (nd) ‘Attracting and Retaining Underprivileged Students’. Washington, DC: World
Bank News & Broadcast. Available online: http://web.worldbank.org/WBSITE/EXTERNAL/
NEWS/0,,contentMDK:23191421~menuPK:141310~pagePK:34370~piPK:34424~theSitePK:4607,00.html.
World Bank (2013) Seeding Fertile Ground: Education that works for Bangladesh. Washington, DC:
World Bank. Available online: http://documents.worldbank.org/curated/en/830371468212071486/
Seeding-fertile-ground-education-that-works-for-Bangladesh

Child labour and education – a survey of slum settlements in Dhaka  75  

ODI is the UK’s leading independent
think tank on international
development and humanitarian
issues.

Readers are encouraged to

reproduce material from ODI
Reports for their own publications,
as long as they are not being sold
commercially. As copyright holder,
ODI requests due acknowledgement
and a copy of the publication. For
online use, we ask readers to link
to the original resource on the
ODI website. The views presented
in this paper are those of the
author(s) and do not necessarily
represent the views of ODI.
© Overseas Development Institute
2016. This work is licensed under
a Creative Commons Attribution-
NonCommercial Licence
(CC BY-NC 4.0).

All ODI Reports are available

from www.odi.org

Cover photo: Abir Abdullah/ADB:

Sohel works in an aluminium
factory at Kamrangirchar in Dhaka.

Overseas Development Institute

203 Blackfriars Road
London SE1 8NJ
Tel +44 (0) 20 7922 0300
Fax +44 (0) 20 7922 0399

This material has been funded by UK aid

from the UK Government, however the views

odi.org expressed do not necessarily reflect the UK

Government’s official policies.